Method for constructing photovoltaic output scenarios in plum rain season based on weibull distribution

Abstract

The application relates to the technical field of photovoltaic output and discloses a plum rain season photovoltaic output scene construction method based on a Weibull distribution, which comprises the following steps: obtaining historical meteorological data, establishing a rainfall intensity sample set and generating a rainfall intensity time sequence matrix; obtaining a non-attenuation reference irradiance based on the clear sky theory, comprehensively considering irradiance influencing factors, and obtaining an irradiance time sequence matrix according to the rainfall intensity time sequence matrix; and obtaining a photovoltaic output probability scene set based on the irradiance time sequence matrix and in combination with environmental temperature conditions. Through the historical meteorological data, a complete data link of "rainfall intensity time sequence scene description-considering comprehensive irradiance attenuation scene conversion of multiple influencing factors-conversion from irradiance and environmental temperature to photovoltaic output scene-obtaining a photovoltaic output probability scene set" is sequentially completed, seamless transmission of the historical meteorological data is realized, and a blank of a plum rain season photovoltaic output special scene construction method is filled.

Description

Technical Field

[0001] This invention relates to the field of new energy output scenarios in power systems, and more specifically, to a method for constructing photovoltaic output scenarios during the rainy season based on the Weber distribution. Background Technology

[0002] With the increasing proportion of new energy sources integrated into the new power system, the installed capacity of photovoltaic power generation continues to grow. However, its output is significantly affected by weather conditions, posing a severe challenge to the supply and demand balance analysis of the power system. In particular, during the plum rain season, continuous rainy weather leads to a significant reduction in irradiance, causing a sharp drop in photovoltaic output and posing a safety hazard to the power system's supply capacity.

[0003] Currently, there are existing studies on methods for generating power system scenarios under the influence of extreme weather. For example, the invention patent application with publication number CN120430578A discloses a method for generating typical regional source-grid-load supply and demand scenarios considering the influence of extreme weather. It identifies key meteorological elements through correlation analysis, uses K-means clustering and a denoising variational autoencoder to generate a weather scenario set, and establishes a geographic information-driven model of extreme weather impact factors. However, it mainly focuses on modeling the propagation characteristics of extreme weather such as high temperatures and strong winds. Its photovoltaic power generation model only uses temperature as a meteorological correction variable, does not involve the impact mechanism of continuous rainy weather during the plum rain season on photovoltaic output, and does not introduce a quantitative model of the attenuation of irradiance due to rainfall. It cannot reflect the drastic change in irradiance during the plum rain season, which drops to 10%-20% when it is sunny.

[0004] For example, patent application CN121791112A discloses a research method for multi-temporal and spatial power supply planning in power systems that considers the impact of extreme weather. It obtains rainfall intensity based on the Weiber distribution and calculates the rainfall attenuation coefficient, then obtains the actual irradiance based on the rainfall attenuation coefficient and Meiyu (plum rain season) irradiance data. However, its rainfall attenuation coefficient is directly expressed as a power function of rainfall intensity, without systematically modeling the probability distribution of rainfall intensity itself, resulting in a lack of statistical basis for obtaining the rainfall intensity sequence. Furthermore, it does not consider the impact of topographical spatial differences on Meiyu irradiance attenuation. It uses the FS statistical method to generate typical meteorological years, which can only output deterministic typical scenarios and cannot generate a set of photovoltaic output probability scenarios covering all possible rainfall conditions, thus failing to adequately characterize extreme low-output scenarios.

[0005] Therefore, existing technologies lack specialized construction methods for photovoltaic power output scenarios during the rainy season, making it impossible to provide high-coverage and high-precision scenario data support for power system supply guarantee analysis and risk assessment during the rainy season. Summary of the Invention

[0006] To overcome the aforementioned deficiencies of the prior art, this invention provides a method for constructing photovoltaic power output scenarios during the rainy season based on the Weber distribution, in order to address the problem that the probability distribution characteristics of rainfall intensity and the spatial differences in irradiance attenuation during the rainy season are not considered, resulting in insufficient characterization of photovoltaic power output scenarios during the rainy season.

[0007] To achieve the above objectives, embodiments of the present invention provide a method for constructing photovoltaic power output scenarios during the rainy season based on the Weibull distribution, comprising: acquiring historical meteorological data, establishing a rainfall intensity sample set and generating a rainfall intensity time series matrix; obtaining a non-attenuated baseline irradiance based on clear sky theory, and comprehensively considering irradiance influencing factors, obtaining an irradiance time series matrix based on the rainfall intensity time series matrix, wherein the irradiance influencing factors include rainfall attenuation, topographic spatial differences and cloud cover effects; and obtaining a photovoltaic power output probability scenario set based on the irradiance time series matrix and combined with ambient temperature conditions.

[0008] First, historical meteorological data for the target area is acquired, and a sample set of rainfall intensity during the plum rain season is established based on the rainfall intensity data during the plum rain season from the historical meteorological data. Then, a time series matrix of hourly rainfall intensity during the plum rain season is generated based on this sample set. This lays the foundation for rainfall intensity data in subsequent physical modeling.

[0009] Then, based on the clear-sky theoretical calculation model, the attenuated baseline irradiance of the target area at any time is obtained; and then, based on the generated hourly rainfall intensity time series matrix of the plum rain season... Taking into account the factors affecting irradiance, a comprehensive irradiance attenuation model is established, which couples multiple influencing factors such as rainfall attenuation, topographic spatial differences, and cloud cover effects. The generated hourly rainfall intensity time series matrix for the plum rain season is then used to analyze this model. Substituting into the comprehensive irradiance attenuation model, an hourly irradiance time series matrix is ​​generated for the rainy season. This enables the physical transformation from rainfall scenarios to irradiance scenarios, and further enables the mapping transformation from the rainfall intensity time series matrix to the irradiance time series matrix.

[0010] Finally, based on the rainfall intensity sequence matrix and combined with the ambient temperature conditions, a temperature time series matrix corresponding one-to-one with the rainfall intensity sequence matrix is ​​generated. Then, based on the conversion method of irradiance and ambient temperature into photovoltaic power output, the irradiance time series matrix is... Corresponding scenario and temperature time series matrix The corresponding scenario is converted into a photovoltaic power output probability scenario to obtain the hourly photovoltaic power output time series matrix during the rainy season. The probability of occurrence of photovoltaic power output scenarios is used to form a complete set of photovoltaic power output probability scenarios.

[0011] The beneficial effects of this invention are: (1) Starting from historical meteorological data, the complete data link is completed in sequence: “temporal scene characterization of rainfall intensity - comprehensive irradiance attenuation scene conversion considering multiple influencing factors such as rainfall attenuation, topographic spatial differences and cloud cover effect - conversion from irradiance and ambient temperature to photovoltaic power output scene - obtaining photovoltaic power output probability scene set”, realizing seamless transmission of historical meteorological data and filling the gap in the method of constructing photovoltaic power output special scene during the plum rain season; (2) By comprehensively considering three completely different physical factors, namely rainfall attenuation, topographic spatial differences and cloud cover effect, we modeled and obtained the corresponding coefficients respectively, and constructed a comprehensive irradiance attenuation model with multiple influencing factors coupled together. This model accurately depicted the superposition of multiple interrelated influencing factors on irradiance and accurately reflected the superposition effect of multiple different physical factors on irradiance. (3) Based on the time series matrix of rainfall intensity and the time series matrix of irradiance, considering the influence of ambient temperature on photovoltaic power output, and combined with ambient temperature conditions, the accuracy of the actual photovoltaic power output scenario is further improved; and by obtaining the photovoltaic power output probability scenario set, an accurate probability scale is provided for power system risk assessment. Attached Figure Description

[0012] Figure 1 A flowchart illustrating the method for constructing photovoltaic power output scenarios during the rainy season based on the Weibull distribution; Figure 2 This is a curve showing the fitting of rainfall intensity data to the Weibull distribution in historical meteorological data. Figure 3 This is a schematic diagram of the Latin hypercube sampling (LHS) process; Figure 4 A schematic diagram of a comprehensive irradiance attenuation model involving the coupling of multiple influencing factors; Figure 5 Example diagram illustrating the probability scenario of photovoltaic power output during the rainy season. Detailed Implementation

[0013] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0014] Please see Figure 1-5 To more clearly illustrate the method for constructing photovoltaic power output scenarios during the rainy season based on the Weibull distribution, the present invention is described in detail below: S1. Obtain historical meteorological data, establish a sample set of rainfall intensity during the plum rain season, and use two-parameter Weiber distribution and Latin hypercube sampling to generate a time series matrix of hourly rainfall intensity during the plum rain season; By statistically analyzing historical meteorological data from the Meiyu season in the target area, a Weiber distribution model capable of accurately characterizing its probability distribution is established. Based on this model, Latin hypercube sampling is used to generate a statistically representative set of rainfall intensity scenarios, obtaining a time-series matrix of hourly rainfall intensity during the Meiyu season. This provides fundamental data support for subsequent irradiance attenuation modeling and photovoltaic power output scenario generation. Specifically: S11. Obtain historical meteorological data, identify the plum rain season periods in the historical meteorological data, and establish a plum rain season rainfall intensity sample set based on the rainfall intensity data of the plum rain season periods. S111. Obtain historical meteorological data: Obtain at least 10 years of historical hourly meteorological observation data for the target area, including... Irradiance at time (unit: ), ambient temperature (unit: Rainfall intensity (Unit: mm / h) and cloud cover (Values ​​range from 0 to 1, where 0 represents clear skies and 1 represents completely cloudy skies, calculated based on total cloud cover data from meteorological observations). Simultaneously, obtain the technical parameters of the target photovoltaic power station, including: rated power. (Unit: MW) Photovoltaic module standard operating condition conversion efficiency (Typically 15%-22%) and power temperature coefficient (Typically -0.005 to -0.003 / ) ).

[0015] Furthermore, the acquired historical meteorological data can be preprocessed. This preprocessing includes: checking for missing values ​​(for single-point missing values, linear interpolation is used to fill them; for periods with consecutive missing values ​​exceeding 6 hours, the data for that day is directly removed, and the number of removed samples is recorded); removing abnormal records that clearly exceed physical limits (such as negative irradiance values, negative rainfall intensity values, and temperatures exceeding the range); and checking the physical consistency between irradiance and cloud cover (when cloud cover...). At that time, the irradiance should not exceed 300. If a record violates this physical constraint, it is marked as suspicious data and manually reviewed. This preprocessing ensures the validity of historical meteorological data.

[0016] S112. Identify the plum rain season periods in historical meteorological data, and establish a rainfall intensity sample set based on the rainfall intensity data of the plum rain season periods; The plum rain season is identified using a dual criterion of daily rainfall and average daily irradiance: when the daily rainfall and average daily irradiance are simultaneously met for 5 consecutive days or more... And daily average irradiance 200 At that time, it was determined to be the plum rain season.

[0017] This dual criterion incorporates daily average irradiance. 200 As an auxiliary criterion, compared with traditional meteorology which relies solely on rainfall or weather phenomenon observations, this method allows the identification results to directly match the physical mechanism of the sharp drop in photovoltaic output during the plum rain season, namely, when the average daily irradiance is below 200 ppm. This means that photovoltaic output will drop to less than 20% of the rated value. This quantitative criterion provides a sample basis directly related to the supply guarantee analysis for subsequent scenario generation.

[0018] Rainfall intensity data for all times (hours) is extracted from the identified plum rain season periods to establish a plum rain season rainfall intensity sample set: In the formula, for The intensity of rainfall at any given moment; The set of all times (hours) labeled as periods of the rainy season, with the total sample size denoted as . At the same time, statistical information such as the annual distribution and time period distribution of the rainfall intensity sample set during the plum rain season will be recorded for reference in subsequent analysis.

[0019] S12. Based on the rainfall intensity sample set during the plum rain season, a two-parameter Weiber distribution is used to fit the probability density function of rainfall intensity, and the scale parameter is obtained through maximum likelihood estimation. and shape parameters ; Please see Figure 2 The probability density of the rainfall intensity sequence is fitted using a two-parameter Weiber distribution, and the probability density function is: , , In the formula, Rainfall intensity (unit: mm / h); This is a scale parameter (unit: mm / h), reflecting the average level of rainfall intensity; The shape parameter (dimensionless) reflects the shape characteristics of the rainfall intensity distribution. Characterizing high-frequency light rain events, The distribution tends to be concentrated.

[0020] By fitting the probability density of the rainfall intensity sequence with a two-parameter Weibull distribution, a complete estimate is obtained. and Two parameters accurately characterize the skewed distribution of rainfall intensity during the plum rain season, which features frequent light rain and low-frequency but significant extreme heavy rainfall.

[0021] The maximum likelihood estimation method is used for the scale parameter and shape parameters Estimates were made based on a sample set of rainfall intensity during the plum rain season. The log-likelihood function is constructed as follows: , Based on this log-likelihood function, respectively for and By taking the partial derivatives and setting them to zero, the Newton-Raphson method is used to solve the nonlinear equations and obtain parameter estimates. and .

[0022] The Newton-Raphson method is used to solve the nonlinear equation system. Compared with the method of moments, it has higher asymptotic efficiency and smaller estimation bias under small sample conditions.

[0023] Furthermore, to verify the fitting effect of the Weiber distribution on the rainfall intensity sample, two testing methods can be used: the KS test and the root mean square error (RMSE) evaluation. When using the KS test, the maximum absolute difference between the empirical distribution function and the fitted Weiber distribution is calculated. At a significance level of α=0.05, if the maximum absolute difference is less than the critical value, the null hypothesis that "the sample follows a Weiber distribution" is accepted. When using the RMSE evaluation, the root mean square error between the empirical quantile and the theoretical quantile at each quantile point is calculated. The smaller the RMSE value, the better the fitting effect, and it is generally required that the RMSE is not greater than 0.05.

[0024] S13. Probability density function of rainfall intensity during the plum rain season obtained based on fitting. and its scale parameters and shape parameters Latin hypercube sampling was used to generate hourly rainfall intensity time series matrices for the plum rain season; S131. Based on the historical statistical characteristics of the target area, set the scene generation parameters including: the duration of the rainy season. (Based on the average number of days of all identified rainy season periods, rounded up) Total hours per scene ( ) and the total number of scenes (Set according to computing resources and accuracy requirements).

[0025] S132. Probability density of rainfall intensity sequence obtained based on fitting during the plum rain season. Latin hypercube sampling was employed, introducing a two-stage sampling strategy of "daily rainfall sampling + hourly allocation" to obtain daily rainfall and its hourly allocation. Please refer to... Figure 3 Specifically: S1321, For each scene ( Sampling from the Weiber distribution Rainfall the following day ( (The duration of the rainy season). Latin hypercube sampling was used to divide the probability intervals. Divided into equal parts Each subinterval contains a probability value, and a probability value is drawn from each subinterval. Daily rainfall is calculated using the inverse cumulative distribution function. for: , , In the formula, is the inverse cumulative distribution function of the Weiber distribution.

[0026] By employing Latin hypercube sampling, it can be ensured that sampling points are uniformly distributed throughout the entire probability space of rainfall intensity, thereby ensuring that extreme heavy rainfall events located at the right tail of the distribution (although the probability is low, they have a huge impact on photovoltaic power output) are inevitably covered by the sampling.

[0027] S1322. Based on historical data, the hourly distribution coefficient is obtained by statistically analyzing the average proportion of hourly rainfall to daily rainfall. ( ),satisfy , will the Daily rainfall The hourly rainfall intensity, distributed across the days, is as follows: , In the formula, This provides a global hour index for the scene. ; For date index ( ); For intraday hourly index ( ).

[0028] In one specific embodiment, to increase the diversity of daily rainfall patterns, the following can be implemented: Introducing random perturbations Then renormalize: After normalization replacement The rainfall patterns should be allocated to avoid all days having the exact same pattern.

[0029] S1323, The above-generated... The scenarios are statistically validated using the first moment (mean), second moment (standard deviation), and quantiles (5%, 50%, 95%), with the relative errors of each statistic not exceeding 5%, 10%, and 10%, respectively. If any validation fails, the number of scenarios is automatically increased. (For example, increase from 1000 to 2000 groups) Regenerate until the verification requirements are met.

[0030] The statistical verification mechanism of this scene set ensures that the generated scenes are strictly consistent with historical measured data in terms of statistical characteristics.

[0031] Finally, the statistical verification will be passed. The rainfall intensity scenarios are output in matrix form (i.e., the hourly rainfall intensity time series matrix for the plum rain season is obtained as follows): , In the formula, The total number of hours during the rainy season will be used as the input for the subsequent integrated irradiance attenuation model.

[0032] In summary, the Latin hypercube sampling (LHS) method is used to generate rainfall intensity scenarios. Through stratified sampling, it ensures the uniform distribution of sampling points in the probability space, effectively covering low-probability extreme heavy rainfall events and significantly improving the scenario set's ability to characterize extreme low-output risks. At the same time, through a two-level sampling strategy of "daily rainfall sampling + hourly allocation" and a statistical verification mechanism, it ensures that the generated scenarios are consistent with historical statistical characteristics on both daily and hourly scales.

[0033] S2. Establish a theoretical irradiance calculation model for clear sky, obtain the baseline irradiance without attenuation, and comprehensively consider the factors affecting irradiance. Based on the hourly rainfall intensity time series matrix of the plum rain season, transform to obtain the hourly irradiance time series matrix of the plum rain season. The factors affecting irradiance include: rainfall attenuation, topographic spatial differences, and cloud cover effect. Please see Figure 4 Considering that the actual irradiance received by photovoltaic panel surfaces during the rainy season is affected by multiple factors: direct scattering and absorption attenuation of solar radiation from rainfall, spatial modulation effect of topography on irradiance, and shading attenuation of solar radiation from cloud cover, a comprehensive irradiance attenuation model coupling three main influencing factors—rainfall attenuation, topography correction, and cloud cover—is constructed. This model converts the hourly rainfall intensity time-series matrix generated by S1 during the rainy season into an irradiance time-series matrix, providing accurate irradiance input for subsequent photovoltaic power output calculations. Specifically: S21. Establish a theoretical irradiance calculation model for clear sky and obtain the unattenuated reference irradiance of the target area at any time. Theoretical clear-sky irradiance refers to the theoretical irradiance of solar radiation reaching the Earth's surface after being absorbed, scattered, and attenuated by the atmosphere. This theoretical irradiance serves as a benchmark value for the effects of various subsequent attenuation factors.

[0034] S211. Based on the geographical latitude of the target area (Unit: radians), declination angle (Unit: radians) Time Angle (Unit: radians) Calculate the solar zenith angle. (Unit: radians) is: , Among them, declination angle Calculated based on date; hour angle Calculated based on local solar time.

[0035] S212, Calculate the intensity of solar radiation outside the atmosphere. (unit: )for: , In the formula, The solar constant (taken as) ); It is the accumulated days of the year (January 1st is 1, December 31st is 365).

[0036] S213. Calculate theoretical clear-sky irradiance using the ASHRAE clear-sky model. (unit: )for: , In the formula, The value is the clear-sky atmospheric transmittance, determined based on the climate characteristics of the target area (typically ranging from 0.6 to 0.8). This represents the atmospheric mass number.

[0037] For tilted photovoltaic modules, the horizontal irradiance needs to be further converted to the tilted irradiance of the module as follows: , In the formula, This is the angle between the solar incidence angle and the normal to the photovoltaic module, calculated based on the module tilt angle and azimuth angle.

[0038] The ASHRAE Clear Sky Model, proposed by the American Society of Heating, Refrigeration and Air-Conditioning Engineers (ASHRAE), is an engineering calculation model used to estimate the intensity of solar radiation reaching the Earth's surface under cloudless (clear sky) conditions. This model has a clear physical mechanism and simple parameter calibration. Using the ASHRAE Clear Sky Model to calculate the theoretical irradiance under clear sky conditions as a benchmark value can fully consider the influence of geographical latitude and seasonal variations of the target area on the theoretical irradiance, providing a reliable benchmark reference for the subsequent quantification of the attenuation coefficient.

[0039] S22. Based on rainfall attenuation factors, considering the scattering and absorption attenuation effects of rainfall on solar radiation, establish a rainfall attenuation coefficient model and obtain the rainfall attenuation coefficient. ; The attenuation of solar radiation by rainfall mainly includes the following two physical processes: (1) the scattering effect of raindrops on solar radiation, which scatters some of the direct radiation to other directions; (2) the absorption effect of raindrops on solar radiation, which converts some of the radiation energy into heat energy.

[0040] The nonlinear relationship between rainfall intensity and attenuation coefficient is characterized using a power function, specifically: S221. The rainfall attenuation coefficient model is established as follows: , In the formula, The rainfall intensity is The rainfall attenuation coefficient at that time has a range of values. ; , This is an empirical coefficient for irradiance attenuation, reflecting the sensitivity of rainfall to radiation attenuation. The upper limit of the maximum attenuation coefficient (take) When, it means that under extreme heavy rainfall conditions, the irradiance retains at least 10% of the theoretical value for sunny days. Its value is based on physical constraints, meaning that even under the strongest rainfall, some scattered radiation still reaches the surface.

[0041] S222. Using the nonlinear least squares method, empirical coefficients are determined based on historical data fitting. and The objective function is constructed as follows: , In the formula, This represents the total sample size for the historical plum rain season. For the first Theoretical clear-sky irradiance for a single sample; For the first The actual measured irradiance of each sample; For the first Rainfall intensity for each sample.

[0042] The Levenberg-Marquardt algorithm (LMA) is used to solve the above nonlinear least squares problem. This algorithm combines the advantages of the Gauss-Newton method and the gradient descent method, and has good convergence in the case of large residuals.

[0043] S223. Set up a physical boundary verification mechanism for the rainfall attenuation coefficient model, and perform the following physical boundary verification on the rainfall attenuation coefficient model: (1) Monotonicity check: Should follow The value increases and then monotonically decreases, that is... From the rainfall attenuation coefficient model formula, it can be seen that when , hour, It is monotonically increasing, therefore its monotonicity is automatically satisfied.

[0044] (2) Saturation check: When hour, The attenuation tends to saturate, which is consistent with physical reality. That is, infinitely increasing the rainfall intensity will not cause the irradiance to infinitely approach zero, and there will always be scattered radiation reaching the earth's surface.

[0045] (3) Boundary condition verification: when hour, There is no attenuation when there is no rainfall, which satisfies the physical boundary conditions.

[0046] In summary, when establishing the rainfall attenuation coefficient model, a physical saturation mechanism is introduced. This avoids the non-physical problem of the attenuation coefficient continuously increasing as rainfall intensity increases indefinitely; the use of nonlinear least squares fitting and LMA solution makes the parameter estimation have a clear optimization objective and a reproducible numerical solution process; by establishing a complete physical boundary verification system, the physical rationality of the model can be ensured across the entire rainfall intensity range.

[0047] S23. Based on spatial differences in terrain, and considering the spatial modulation effect of altitude and surface roughness on irradiance, establish a terrain correction coefficient model and obtain the terrain correction coefficient. ; The influence of topography on surface irradiance is mainly reflected in two aspects: (1) the higher the altitude, the shorter the atmospheric path, the smaller the atmospheric attenuation, and the relatively larger the surface irradiance; (2) the surface roughness affects the surface reflectivity and local microclimate, which in turn affects the effective irradiance reaching the surface of photovoltaic modules.

[0048] By acquiring topographic and geomorphological information of the target area, a topographic correction coefficient model is established, and the topographic correction coefficients are obtained. Specifically: S231. Obtain the topographic information of the target area, including: altitude. (Unit: m, calculated based on digital elevation model), Surface roughness coefficient (Dimensionless, range 0.05-0.30, smaller values ​​for plains areas, larger values ​​for mountainous areas, determined based on land use type data), Topographic slope angle (Unit: radians, calculated based on digital elevation model).

[0049] S232. Establish the terrain correction coefficient model as follows: , In the formula, This is the terrain correction factor (dimensionless). For reference altitude, this embodiment takes ,make Under typical terrain conditions, the value falls within a reasonable range of 0.1-10; This is the surface roughness coefficient, reflecting the influence of surface characteristics on radiation transmission; The slope angle of the terrain. This reflects the correction of slope to effective receiving area; that is, the steeper the slope, the less radiation is received per unit horizontal area.

[0050] S233. Set up a physical boundary verification mechanism for the terrain correction coefficient model, and perform the following physical boundary verification on the terrain correction coefficient: (1) Baseline condition verification: when , , (Sea level, smooth flat ground) There is no terrain correction effect.

[0051] (2) Monotonicity check: Follow Increases and monotonically increases ( This aligns with the physical law that atmospheric attenuation decreases with increasing altitude.

[0052] (3) Value range verification: Under typical terrain conditions ( m, , , The correction range does not exceed ±35%, which is consistent with physical reality.

[0053] In summary, by introducing terrain correction coefficients The comprehensive characterization of the three factors of altitude, surface roughness, and slope enables the terrain correction coefficient model to reflect the spatial differences in irradiance attenuation in different geographical regions during the rainy season. That is, plain areas with low altitude and low roughness and mountainous areas with high altitude and high roughness will receive different actual irradiance under the same rainfall intensity, providing spatial resolution for regional refined photovoltaic power output assessment.

[0054] S24. Based on the cloud cover effect, considering the shading and attenuation effect of clouds on solar radiation, establish a cloud cover attenuation coefficient model and obtain the cloud cover attenuation coefficient. ; The persistent rainy weather during the plum rain season is accompanied by high cloud cover, and cloud obstruction of solar radiation is one of the main reasons for irradiance attenuation. This study establishes a coupled model of cloud cover and rainfall intensity to calculate the cloud cover attenuation coefficient. Specifically: S241. Establish a logistic regression coupling model between cloud cover and rainfall intensity. This model is based on historical meteorological data from the plum rain season regarding rainfall intensity. With cloud coverage The corresponding relationships are fitted using logistic regression as follows: , In the formula, Cloud cover (0 indicates clear skies, 1 indicates completely overcast skies); Rainfall intensity (unit: mm / h); , The parameters for logistic regression are determined by fitting historical meteorological data using maximum likelihood estimation.

[0055] S242. The cloud cover attenuation coefficient model is established as follows: .

[0056] In the formula, This represents the cloud attenuation efficiency coefficient.

[0057] S243. Set up a physical boundary verification mechanism for the cloud cover attenuation coefficient model, and perform the following physical boundary verification on the cloud cover attenuation coefficient model: (1) Clear sky boundary: when hour, ,like Take the larger negative value (e.g.) ),but The sky was almost clear and cloudless. The attenuation is very small.

[0058] (2) Rainy boundary: when hour, , That is, the irradiance retention ratio on a completely cloudy day is It is usually 20%-40%.

[0059] (3) Monotonicity check: Follow Monotonically increasing ( ), Follow The monotonically decreasing trend aligns with the physical law that the greater the rainfall, the thicker the cloud layer and the greater the radiation attenuation.

[0060] In summary, by establishing a logistic regression coupling between cloud coverage and rainfall intensity... This ensures that cloud cover is no longer an independent external input, but rather is generated driven by rainfall intensity scenarios, guaranteeing physical consistency among meteorological elements; the logistic regression model is concise in form. Baseline cloud cover reflecting the amount of cloud cover when there is no rain. It reflects the sensitivity of rainfall to cloud cover increase; by using the maximum likelihood estimation of historical data to fit parameters, the cloud cover attenuation coefficient model has a clear data-driven basis.

[0061] S25, Based on rainfall attenuation coefficient Terrain correction factor and cloud cover attenuation coefficient A comprehensive irradiance attenuation model with multiple influencing factors was established, and the hourly irradiance time series matrix of the plum rain season was obtained by transforming it based on the hourly rainfall intensity time series matrix of the plum rain season. S251, Rainfall attenuation coefficient Terrain correction factor and cloud cover attenuation coefficient When modeling different physical processes independently, the parameters of each model can be independently calibrated based on different types of historical data, without interference. Then, the obtained rainfall attenuation coefficients are... Terrain correction factor and cloud cover attenuation coefficient A comprehensive irradiance attenuation model, characterized by product coupling, is established to represent the combined effect of multiple influencing factors leading to a sudden drop in irradiance: , In the formula, The rainfall intensity is At that time, the actual irradiance reaching the surface of the photovoltaic module (unit: ); The theoretical clear-sky irradiance calculated in S21 (unit: ).

[0062] S252. The comprehensive irradiance attenuation model is validated under the following extreme scenarios: (1) Scene of clear sky and no rain ( ): , It is only affected by topography and a very small amount of cloud cover.

[0063] (2) Moderate rainfall scenario ): with typical parameters , For example, , , ; This means that the actual irradiance drops to about 19% of the theoretical value under clear skies, which is highly consistent with the measured phenomenon of a sharp drop in photovoltaic output during the rainy season.

[0064] (3) Extreme heavy rainfall scenarios ):when hour, ;at the same time , ; That is, the actual irradiance can be reduced to about 3% of the theoretical value under clear skies.

[0065] After validating the comprehensive irradiance attenuation model under extreme scenarios, the hourly rainfall intensity time series matrix generated in S1 for the plum rain season is used. Substituting the values ​​into the comprehensive irradiance attenuation model formula at each time step, the following is generated: An irradiance scene set for the plum rain season was created.

[0066] S253. To further ensure the rationality of the final hourly irradiance scene obtained during the plum rain season, statistical verification and rationality validation can be selectively performed on the generated irradiance scene set. Specifically: S2531, Based on historical measured irradiance data from the plum rain season, the generated... Statistical verification was performed on the irradiance scenarios; the statistical verification included: mean verification, quantile verification, and daily total verification. The mean verification involved calculating... The average irradiance at all times in the irradiance scenario is compared with the historical sample mean, and the relative error is required. The quantile check is performed by calculating... The 5%, 50%, and 95th percentiles of irradiance in the irradiance scenario are compared with the corresponding quantiles in historical samples, and the relative errors of each quantile are all... The daily total verification is as follows: Calculation The relative error of the daily average cumulative irradiance in the irradiance scenario compared with historical samples. .

[0067] S2532, For the generated The rationality of the irradiance scenarios was verified. This verification included: verification of diurnal variation patterns, verification of the negative correlation between rainfall and irradiance, and verification of extreme scenarios. Specifically, the verification of diurnal variation patterns involved checking whether the diurnal variation curves of irradiance in each irradiance scenario exhibited a physically reasonable pattern of "rising after sunrise, reaching a peak at noon, and declining after sunset," and whether there were non-physical phenomena such as noon irradiance being lower than that in the morning or evening. The verification of the negative correlation between rainfall and irradiance involved checking the rainfall intensity in each irradiance scenario. With irradiance The correlation coefficient between the two is negative to ensure that the model correctly reflects the physical law that "the greater the rainfall, the lower the irradiance". The extreme scenario verification is as follows: for the 5% scenario with the greatest rainfall intensity, check whether its irradiance is significantly lower than the average value to ensure that the extreme rainfall scenario produces the expected extreme low irradiance effect.

[0068] If any of the above rationality verifications fails, then check the empirical coefficient of rainfall attenuation in S22 above. , The fitting results, and / or the logistic regression parameters in S24 above. , The fitting results are used to recalibrate the parameters until the verification is successful.

[0069] S254. Actual irradiance reaching the surface of the photovoltaic module obtained based on the comprehensive irradiance attenuation model. Based on the hourly rainfall intensity time-series matrix of the plum rain season, the final hourly irradiance time-series matrix of the plum rain season is obtained by transformation. Specifically: The hourly rainfall intensity time series matrix generated by S1 during the plum rain season Substituting the values ​​into the comprehensive irradiance attenuation model formula at each time step, the following is generated: An hourly irradiance scene during the rainy season. This generated... The hourly irradiance scenarios during the plum rain season were subjected to statistical verification and rationality validation of S253 with or without it, resulting in the final hourly irradiance time series matrix for the plum rain season: , In the formula, , for the first The first scene in the group Actual irradiance at time (unit: ).

[0070] In summary, by establishing independent physical models and parameter calibration methods for the three main attenuation factors affecting irradiance, and verifying and calibrating them based on different types of historical data, and by coupling the three main factors affecting irradiance in a product form, the composite effect of the superposition of multiple factors such as rainfall, cloud cover, and topography during the plum rain season leading to a sharp drop in irradiance is reflected. Specifically, rainfall directly attenuates radiation, clouds block radiation, and topography modulates the spatial distribution of radiation; the superposition of these three factors allows the actual irradiance to be reduced to 3%-10% of the theoretical value under clear skies. Finally, based on a rainfall intensity scenario set, the comprehensive irradiance attenuation model is converted into an irradiance scenario set, establishing a complete mapping from the probability distribution of rainfall intensity to irradiance attenuation scenarios, providing accurate irradiance input for subsequent photovoltaic power output calculations.

[0071] S3. Based on the hourly irradiance time series matrix during the plum rain season, combined with the ambient temperature conditions, obtain the photovoltaic power output probability scenario set and the probability of occurrence of each scenario during the plum rain season. Please see Figure 5 Based on the hourly irradiance time-series matrix of the plum rain season in S2, a temperature time-series generation model driven by rainfall intensity is established to obtain a temperature scenario consistent with the rainfall intensity time-series scenario. Then, the irradiance scenario and the temperature scenario are substituted into the photovoltaic power conversion model to generate photovoltaic power output scenarios. Finally, based on the Weiber distribution probability density of the plum rain season rainfall intensity sequence in S1, the occurrence probability of each group of photovoltaic power output scenarios is calculated to obtain a complete set of probabilistic photovoltaic power output scenarios for the plum rain season and their occurrence probabilities. Specifically: S31. Based on historical meteorological data, establish a temperature time series generation model driven by rainfall intensity, and generate a temperature time series matrix corresponding to the rainfall intensity time series matrix; Ambient temperature during the rainy season significantly impacts photovoltaic (PV) module output (increased temperature leads to decreased photoelectric conversion efficiency), and a correlation exists between temperature and rainfall intensity (temperatures during cloudy / rainy weather are typically lower than those during sunny weather). To ensure consistency between the temperature scenario and the rainfall intensity time series scenario obtained in S1, a conditional autoregressive temperature generation model driven by rainfall intensity is established. Specifically, this includes: S311. Based on the hourly temperature sequence and rainfall intensity sequence in the historical plum rain season meteorological data, divide the rainfall intensity intervals and extract the interval temperature parameters; wherein, the interval temperature parameters include the temperature mean, temperature standard deviation and first-order autocorrelation coefficient.

[0072] Based on hourly temperature series from historical plum rain season meteorological data obtained in S1 (unit: ) and rainfall intensity sequence (Unit: mm / h), rainfall intensity is divided into A range.

[0073] Please refer to Table 1 below. In one specific embodiment, the rainfall intensity is divided into four intervals, namely... .

[0074] Table 1: Statistical Table of Rainfall Intensity Classification

[0075] For each interval Calculate the average temperature within this interval. Temperature standard deviation and first-order autocorrelation coefficient The calculation formulas for the three parameters are as follows: Average temperature , , ; Temperature standard deviation ; First-order autocorrelation coefficient , In the formula, The start time of each rainy season period.

[0076] If the sample size of a certain interval Then, merge the interval with the adjacent interval, or use the global autocorrelation coefficient. Replacement.

[0077] S312. Based on the divided rainfall intensity intervals and the extracted interval temperature parameters, establish a conditional autoregressive model; Given the current Rainfall intensity at any time The interval Temperature at the previous moment Current temperature The conditional distribution adopts the following stationary first-order autoregressive form: , Equivalent land can be expressed as: , in, It is independent Gaussian white noise.

[0078] From this conditional autoregressive model, it can be seen that: when the first-order autocorrelation coefficient... As the temperature approaches 1, the temperature series becomes smooth, conforming to the physical law of continuous temperature change; the conditional mean revolves around the temperature mean. Fluctuations reflect the differences in temperature levels under different rainfall intensities (temperatures are lower on cloudy and rainy days); (3) Conditional variance Ensure that the temperature series statistically converges to an interval steady-state distribution.

[0079] S313. Extract global baseline temperature parameters based on a conditional autoregressive model; For the first hour of the plum rain season ( Temperature initialization was performed, and temperature samples from the first hour of all historical rainy season periods were extracted to calculate the global mean. and global standard deviation : , , In the formula, This is a collection of the first moments of all historical rainy season periods. .

[0080] S32. Based on the hourly rainfall intensity time series matrix and global reference temperature parameters during the plum rain season, sample the time series temperature at each moment and generate a temperature time series matrix. For each rainfall intensity time series scene generated in S1 ( Its time-series rainfall intensity sequence is ,in, This represents the total number of hours during the plum rain season. Based on the hourly rainfall intensity sequence during the plum rain season, a corresponding temperature time series is generated through initial temperature sampling and time-by-time recursive sampling. Specifically: S321. Based on the rainfall intensity time series matrix and global reference temperature parameters, the temperature of the first hour is obtained by independent sampling; Independent sampling generates the first The temperature for the first hour in each scenario is: , To ensure that the sampled values ​​are within a physically reasonable range, if or If the value is not found, resample, with a maximum of 10 retries; if the value is still outside the range, then take the next sample. .

[0081] S322. Based on the rainfall intensity time series matrix and global reference temperature parameters, the temperature at each time point except the first hour is obtained by recursively sampling. for First, based on the current rainfall intensity Determine the interval number Then extract the interval temperature parameters for the corresponding interval. , as well as The conditional mean and conditional standard deviation are calculated as follows: , ; Then, the temperatures at each time point other than the initial time are obtained by sampling from the normal distribution: , And perform physical boundary truncation on the temperature at each time point: if Then set to ;like Then set to .

[0082] S323. Generate a temperature time series matrix based on the initial temperature and the temperatures at each time other than the initial temperature. The generated The temperature scene is output in matrix form (i.e., the temperature time series matrix is ​​obtained) as follows: , In the formula, Indicates the first In the group temperature scenario, the first Ambient temperature at any time (unit: ).

[0083] S33. Establish a conversion model from actual irradiance and ambient temperature to photovoltaic output, and generate a photovoltaic output time series matrix based on the irradiance time series matrix and the temperature time series matrix. Since the real-time output of photovoltaic modules is determined by both irradiance and ambient temperature, a simplified engineering approach to photovoltaic output based on a standard single-diode model is adopted to establish a model from actual irradiance and ambient temperature to photovoltaic output. The conversion model (unit: MW) is as follows: S331. Establish a conversion model from actual irradiance and ambient temperature to photovoltaic power output, wherein the photovoltaic power output conversion model includes a photovoltaic power output conversion formula and photovoltaic power output boundary constraints; S3311, Establish From the actual irradiance (unit: ) and ambient temperature (unit: ) to photovoltaic power output The conversion formula (unit: MW) is as follows: , In the formula, The rated power of the photovoltaic power station (unit: MW) is given by the technical parameters of the photovoltaic power station. The conversion efficiency of photovoltaic modules under standard test conditions (dimensionless, typically 15%-22%). , which is the irradiance under standard test conditions; , which is the battery temperature under standard test conditions; Power temperature coefficient (unit: ), which are usually negative values ​​(e.g., -0.003 to -0.005). ), which represents the percentage decrease in photovoltaic efficiency as temperature increases.

[0084] in, This is a temperature correction term. During the rainy season, the ambient temperature is typically between [temperature range missing]. Between, above the standard test temperature When the correction factor is less than 1, the output decreases; when it is lower than 1, the output decreases. When the correction factor is greater than 1, the output power increases slightly. This correction amount is generally within... Within the range, compared to irradiance from approximately [missing information] on a sunny day The sudden drop to the rainy season (The reduction is as high as 80%-95%). Temperature correction is a minor factor, but its inclusion in the conversion formula can improve the accuracy of photovoltaic output calculation.

[0085] S3312. Establish photovoltaic power output boundary constraints; Physical boundary constraints are applied to the calculated photovoltaic power output, including upper and lower bound constraints.

[0086] The lower bound constraint is: if (Only when the temperature is extremely high and) (This may occur when the absolute value is large), then forced The upper limit constraint is Then forced (To prevent over-generation, the actual inverter has a limiting function.)

[0087] S332. Based on the irradiance time series matrix and the temperature time series matrix, obtain the photovoltaic power output scenario through the photovoltaic power conversion model, and generate the photovoltaic power output time series matrix. The irradiance time series matrix generated in S2 With the temperature time series matrix generated in S32 Substituting each element into the photovoltaic power conversion formula, we can generate... Solar power output scenarios. Specifically: For each scenario ( ) and every moment ( Using the photovoltaic power conversion formula in S331, the first... Group scene, first Photovoltaic output at all times for: .

[0088] The calculated The photovoltaic power output scenario is output in matrix form (i.e., the photovoltaic power output time series matrix is ​​obtained) as follows: , In the formula, For the first The first scene in the group Photovoltaic output per hour (unit: MW).

[0089] In one specific embodiment, rapid verification can also be performed on each photovoltaic output scenario in the photovoltaic output time series matrix. The rapid verification includes daily variation pattern verification, output-irradiance consistency verification, and daily maximum output verification.

[0090] The daily variation pattern verification refers to checking whether the daily output power in each scenario exhibits a single-peak pattern of "rising after sunrise, reaching its peak at noon, and returning to zero after sunset," eliminating abnormal scenarios that violate physical laws, such as output power at noon being lower than in the morning and evening; the output-irradiance consistency verification refers to checking the output power at each time point. With irradiance Is there a positive correlation? Follow (Monotonically increasing); if an anomaly occurs where the irradiance is high but the output is low, mark it and check whether the temperature data is abnormal; the daily maximum output verification: calculate the daily maximum output and check whether it exceeds... (Consider the efficiency improvement when the temperature is below STC), and scenarios that exceed this limit are marked as suspicious but retained.

[0091] If the proportion of scenarios failing the above quick verification exceeds 5%, then the parameters generated by the irradiance time series matrix in S2 (such as the rainfall attenuation coefficient) need to be checked. , Logistic regression parameters , Is this reasonable?

[0092] S34. Based on the photovoltaic power output time series matrix and the joint probability density of daily rainfall under the Weiber distribution, calculate the probability of photovoltaic power output scenarios to obtain the photovoltaic power output probability scenario set during the plum rain season. Because S1 employs a two-level sampling strategy of "daily rainfall sampling + hourly allocation", each scenario Daily rainfall sequence The daily rainfall in the data independently follows a Weiber distribution. The data was obtained through Latin hypercube sampling, therefore the probability of each photovoltaic output scenario occurring is proportional to the product of its daily rainfall probability density. Specifically: S341. Based on the photovoltaic power output time series matrix and the joint probability density of daily rainfall under the Weibull distribution, calculate the... Probability of occurrence of photovoltaic power output scenarios for: , , In the formula, The two-parameter Weiber probability density function fitted in S1 ; and These are the scale parameter and shape parameter obtained by maximum likelihood estimation in S1, respectively.

[0093] In the formula for calculating the probability of a photovoltaic power output scenario, the denominator is all The sum of the combined density of photovoltaic power output scenarios meets the requirements. .

[0094] It should be noted that the probability of a solar power output scenario with zero rainfall is: when the rainfall on a certain day... When, the probability density of the Weiber distribution Note the following when calculating shape parameters: hour, In this case, it is not appropriate to directly substitute the Weiber distribution probability density function, but the following processing strategy can be adopted: when sampling in S1, the daily rainfall is strictly greater than 0 (i.e., for...). (The probability quality is handled separately), the probability interval for Latin hypercube sampling is... It never drops below midnight; in probability calculations, if the rainfall on a certain day... (Pick ), then use replace This ensures numerical stability. Furthermore, this processing strategy only affects the absolute value of the probability; since the denominator uses the same substitution rule, the normalized probability value remains unaffected.

[0095] S342. Integrate the photovoltaic output time series of each scenario with the corresponding occurrence probability to form the final photovoltaic output probability scenario set for the rainy season. Specifically: Define the probabilistic scenario set as: , In the formula, For the first The first scene in the group Photovoltaic output per hour (unit: MW). ; For the first The probability of a group of scenarios occurring.

[0096] Meanwhile, to facilitate subsequent consistency time series verification and statistical feature extraction, the rainfall intensity time series matrix corresponding to each scenario is saved and stored. (Originated from S1) Irradiance time series matrix corresponding to each scene (Originated from S2) Temperature time series matrix corresponding to each scenario (Obtained from S32) and the probability vectors for each scenario. .

[0097] In summary, based on the synchronous observation information of temperature and rainfall from historical meteorological data collected in S1, a conditional autoregressive temperature model based on rainfall intensity intervals ensures physical consistency between temperature and rainfall scenarios (lower temperatures on cloudy / rainy days), avoiding the temperature-rain mismatch problem caused by independent sampling. Based on the statistical characteristics of independently sampled daily rainfall in S1, the probability of occurrence for each complete scenario is calculated using the joint density of the Weipah distribution. This probability reflects the statistical typicality of the daily rainfall sequence; that is, the closer the daily rainfall is to the mode of the distribution, the higher the probability of the scenario, and the lower the probability of the scenario containing extreme rainfall days. This provides an accurate probabilistic benchmark for subsequent power system risk assessment. Furthermore, a temperature correction term is introduced into the photovoltaic power conversion formula. This improves the accuracy of photovoltaic power output calculation.

[0098] S4, Scenario set for photovoltaic power output probability during the rainy season Perform timing consistency checks to obtain the optimal photovoltaic power output probability scenario set during the rainy season. Statistical characteristic set of photovoltaic power output during the plum rain season; Based on the probability scenario set of photovoltaic output during the rainy season generated in S3 (Include Solar power output timing scenarios and the probability of occurrence of each scenario By introducing temporal consistency physical constraints on the scene set Verification and screening are conducted to ensure that all scenarios have reasonable physical continuity in the time dimension. Based on this, the expected output value, standard deviation, and multi-quantile boundaries are calculated for each time step to extract typical statistical characteristics of photovoltaic output during the rainy season. Furthermore, extreme low-output scenarios are identified, and their total probability and expected duration are calculated, providing comprehensive decision-making basis for power system reserve capacity allocation, dispatch planning, and supply guarantee risk assessment during the rainy season. Specifically: S41. Construct time-series consistency constraints and apply them to the photovoltaic power output probability scenario set during the rainy season. Perform timing consistency checks and resampling to obtain the optimal photovoltaic power output probability scenario set for the rainy season. ; Since photovoltaic (PV) output is driven by changes in solar irradiance, abrupt changes exceeding physical limits should not occur between adjacent moments (hours). Although rapid cloud movement during the rainy season may cause short-term fluctuations in irradiance, the rate of change in PV output still has a reasonable upper limit due to the inertia of the PV system and the response speed of the inverter. Furthermore, the PV output curve should satisfy the basic physical form of diurnal variation and should not exhibit non-physical phenomena such as midday irradiance being lower than that in the morning or evening. Based on this, the following time-series consistency constraints are established, and a set of probabilistic scenarios for PV output during the rainy season is applied based on these constraints. By performing verification screening and resampling, the optimal photovoltaic power output probability scenario set for the rainy season was obtained. .

[0099] S411. Construct time-series consistency constraints, which include constraints on the rate of change of output at adjacent times and constraints on the smoothness of output per day. S4111. Establish constraints on the rate of change of output at adjacent time points; For any scenario ( and any adjacent time ( (Total hours during the rainy season), its output variation rate should satisfy: , In the formula, This represents the maximum permissible rate of change (in MW / h) of photovoltaic power output between adjacent time periods (hours). In one specific embodiment, we take... This means that the change in photovoltaic output within one hour should not exceed 30% of the rated capacity. The threshold is determined based on the installed capacity of photovoltaic power plants and engineering experience. It can be verified by taking the 95th percentile based on the statistical analysis of the measured output change rate on typical days during the rainy season.

[0100] S4112. Establish daily power smoothness constraints; To identify spikes and glitches in the output curve (such as non-physical abrupt changes caused by abnormal fluctuations in irradiance at certain times), a three-point sliding smoothness constraint (i.e., sunrise output smoothness constraint) is introduced as follows: , , In the formula, This is the allowable limit for peak deviation. In one specific embodiment, it is taken as... This means that the deviation of the output from the average value of the preceding and following times at any given moment shall not exceed 15% of the rated capacity. The daily output smoothness constraint can effectively eliminate non-physical output spikes caused by abnormal fluctuations at individual moments in the irradiance scenario.

[0101] S412. Based on the time-series consistency constraint, the photovoltaic power output probability scenario set during the rainy season is analyzed. Perform group-by-group verification and screening; Based on S3 For each photovoltaic power output timing scenario, perform constraint checks S411-S412 on each group. If a certain scenario... If any constraint condition is violated at any time, either the constraint on the rate of change of output between adjacent times or the constraint on the smoothness of daily output, then that scenario is marked as an anomalous scenario and removed from the scenario set. Let the number of anomalous scenarios removed be... The remaining number of scenarios that passed verification is The remaining verified photovoltaic power output probability scenario set for the rainy season is as follows: .

[0102] S413. Based on the verified and filtered photovoltaic output probability scenario set for the rainy season and its number of scenarios, resampling is performed to supplement the optimal photovoltaic output probability scenario set for the rainy season. ; To ensure that the final scene set maintains a fixed number of scenes To facilitate a unified format for subsequent statistical analysis, from those that have passed validation. In group scenarios, according to probability Perform resampling with replacement to supplement Group. Specifically: S4131. Renormalize the probability of the scenarios that passed the verification: , ; S4132, from According to probability Extraction with replacement The resampled scenes are used to replace the rejected abnormal scenes; and the resampled scenes are merged with the original validated scenes to form a new set. Scene set This ensures that all scenarios pass the timing consistency check.

[0103] Furthermore, the statistical information of this verification filtering and resampling supplementation can be recorded and output, including: the total number of original scenes. , through the verification of scene numbers Number of abnormal scenarios Validation pass rate .like If the error occurs, a warning will be issued, prompting the need to check the parameter calibration of the irradiance attenuation model in S2 (such as the empirical coefficient of rainfall attenuation). , The fitting results and logistic regression parameters , The fitting results (or random seeds from Latin hypercube sampling in S1) are used to regenerate the scene and then validate it again.

[0104] S42, Scenario Set Based on Optimal Photovoltaic Output Probability During the Plum Rain Season The photovoltaic power output statistical characteristics at each time step are calculated; the photovoltaic power output statistical characteristics at each time step include the expected value of power output at each time step, the standard deviation of power output at each time step, and the quantile of power output at each time step. Based on the optimal solar power output probability scenario set obtained in S41 during the rainy season Calculate the hourly statistical characteristics of photovoltaic power output during the plum rain season: S421, Scenario Set Based on Optimal Photovoltaic Output Probability During the Plum Rain Season The expected output at each time step is calculated as follows: , , The expected value curve of the hourly output reflects the average level of photovoltaic output at each moment during the rainy season and can be used as a benchmark reference for the formulation of power system dispatch plans.

[0105] S422, Scenarios Based on Optimal Photovoltaic Output Probability During the Plum Rain Season The standard deviation of output at each time step is calculated as follows: , , The time-by-time output standard deviation This characterizes the dispersion of power output around the expected value at different times, reflecting the magnitude of uncertainty in photovoltaic power output during the rainy season. Standard deviation The larger the value, the more significantly the power output is affected by fluctuations in rainfall intensity at that moment, and the more flexible adjustment capacity the power system needs to reserve.

[0106] S423, Scenarios Based on Optimal Photovoltaic Output Probability During the Plum Rain Season Calculate the output quantiles at each time step; For a given quantile level quantile of output at each moment Defined as: , In the formula, This is an indicator function.

[0107] In one specific embodiment, the time-by-time values ​​of the following five quantiles can be calculated: (Low output boundary) (Lower quartiles) (Median, i.e., typical output level) (Upper quartiles) and (High output boundary). This quantile band (5%-95%) depicts the main fluctuation range of photovoltaic output during the rainy season and can be used to quantitatively assess the system operation risk under different output levels.

[0108] S43. Based on the time-by-time photovoltaic power output statistical characteristics, identify extreme low power output scenarios and calculate their total probability and expected duration. The persistent rainy weather during the plum rain season leads to extremely low photovoltaic (PV) output for extended periods, posing a key risk factor to the security of the power system. This study identifies extreme low-output scenarios from the optimal plum rain season PV output probability scenario set, extracting features such as the total probability and expected duration of these extreme low-output scenarios to provide directly usable decision-making data for power system dispatching and planning. Specifically: S431. Define and identify extreme low-output scenarios; Define an extreme low-output event (i.e., an extreme low-output scenario) as: in any continuous Within one hour, the photovoltaic output never exceeds the extreme low output threshold coefficient of the rated power. .

[0109] In one specific embodiment, the extreme low output threshold coefficient is taken as... (Right now Take the minimum continuous duration as Hours (i.e., at least 6 consecutive hours in an extremely low output state).

[0110] For the scenario If there exists at least one continuous time period (length) (hours) to satisfy If the scenario is not specified, it will be marked as an extreme low-output scenario and added to the extreme scenario set. : .

[0111] S432. Based on the identified extreme low-output scenarios, calculate the total probability and expected duration of the extreme low-output scenarios. The total probability of an extremely low output scenario is: .

[0112] The total probability of this extreme low-output scenario quantifies the likelihood of a sustained period of extreme low output during the rainy season and is a core indicator for supply guarantee risk assessment.

[0113] For each extremely low output scenario Identify all its extreme low-output continuous periods and record its longest continuous duration as And the total number of extreme low output hours is .

[0114] Further calculations of the expected duration and total expected hours for extreme low output are as follows: , .

[0115] S433. Based on the total probability and expected duration of the extreme low output scenario, obtain the minimum output value and its probability of occurrence of the global minimum output scenario. Identify the instantaneous lowest output scenarios and their probability of occurrence throughout the entire rainy season from all scenarios (or identified extreme low output scenarios): , That is, the minimum output value of this global minimum output scenario is The probability of this scenario with the lowest global output is .

[0116] Minimum output value for this global minimum output scenario and the probability of its occurrence is It can provide a reference for predicting the minimum output in extreme power supply scenarios.

[0117] S44. Based on the time-by-time statistical characteristics of photovoltaic power output, extreme low power output scenarios and their total probability and expected duration, obtain the statistical characteristic set of photovoltaic power output during the plum rain season; By integrating the time-by-time photovoltaic output statistical characteristics obtained from S42 calculation and the extreme low output scenarios and their total probability and duration obtained from S43, a complete statistical characteristic set of photovoltaic output during the rainy season is obtained. This statistical characteristic set provides a decision-making basis for power system dispatching and planning. Specifically: S441. Based on the time-by-time statistical characteristics of photovoltaic power output, the time-by-time statistical characteristic dataset of the plum rain season is obtained as follows: , The statistical characteristic dataset of this rainy season at each moment can be imported into the power system's dispatch platform as a reference for the prediction range and uncertainty of photovoltaic output at each moment during the rainy season.

[0118] S442. Based on extreme low-output scenarios, their total probability, and expected duration, the set of extreme low-output scenarios and their characteristics are summarized as follows: , The overall statistical indicators for extremely low output are: .

[0119] Statistical feature dataset based on the time-by-time characteristics of the plum rain season Summary of Extremely Low Output Scenarios and Their Characteristics and overall statistical indicators of extremely low output This allows for the extraction of key decision support indicators for power system dispatching and planning. For example, in one specific embodiment, reserve capacity demand indicators can be extracted. (That is, the increased reserve capacity required due to the uncertainty of photovoltaic output when the reliability level is 95%), and identify The lowest continuous 6-hour period is the critical period when the pressure to ensure supply is greatest during the plum rain season.

[0120] In summary, by constructing a multi-dimensional temporal consistency constraint verification system that includes constraints on the rate of change between adjacent time points and three-point sliding smoothness constraints, the physical rationality of the generated scenes is ensured at different time scales, effectively avoiding the non-physical mutation problems common in traditional scene generation methods; based on the occurrence probabilities of each scene obtained by S3, the system achieves this. The expected value, standard deviation, and quantiles are calculated using probability weighting, so that the statistical characteristics not only reflect the numerical distribution of the scene set, but also their actual frequency of occurrence in long-term climate statistics; continuous duration is introduced. Using hours as a condition for identifying extremely low-output scenarios avoids misjudging instantaneous low output as extreme events, while also defining the total probability. Expected longest duration Expected total low output hours Multiple indicators comprehensively depict the intensity, duration, and frequency of extreme low-output risks; finally, reserve capacity requirements are extracted. It is used for power system dispatching and planning decisions during extreme risk periods, condensing massive amounts of scenario data into compact and operable decision information, forming a complete technical route closed loop of "probabilistic modeling - scenario generation - scenario verification - statistical extraction - decision support", filling the gap in the connection between the generation and engineering application of photovoltaic power output scenarios during the rainy season.

[0121] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0122] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0123] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0124] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0125] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0126] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for constructing photovoltaic power output scenarios during the rainy season based on Weiper distribution, characterized in that, include: Acquire historical meteorological data, establish a rainfall intensity sample set, and generate a rainfall intensity time series matrix; The baseline irradiance without attenuation is obtained based on the clear sky theory. Taking into account the factors affecting irradiance, the irradiance time series matrix is ​​obtained based on the rainfall intensity time series matrix. The factors affecting irradiance include rainfall attenuation, topographic spatial differences and cloud cover effect. Based on the irradiance time series matrix and combined with ambient temperature conditions, a set of photovoltaic power output probability scenarios is obtained.

2. The method for constructing photovoltaic power output scenarios during the rainy season based on Weipah distribution according to claim 1, characterized in that, The process of establishing a rainfall intensity sample set and generating a rainfall intensity time series matrix includes: Identify the plum rain season periods in historical meteorological data, and establish a rainfall intensity sample set based on the rainfall intensity data of the plum rain season periods; Based on the rainfall intensity sample set, a two-parameter Weiber distribution is used to fit the rainfall intensity probability density function, and the scale parameter and shape parameter are obtained through maximum likelihood estimation. Based on the probability density function of rainfall intensity and its scale and shape parameters, Latin hypercube sampling is used to generate the hourly rainfall intensity time series matrix for the plum rain season.

3. The method for constructing photovoltaic power output scenarios during the rainy season based on Weiper distribution according to claim 2, characterized in that, When identifying the plum rain season period in historical meteorological data, a dual judgment criterion is adopted; when using Latin hypercube sampling, a two-level sampling strategy of "daily rainfall sampling + hourly allocation" is introduced.

4. The method for constructing photovoltaic power output scenarios during the rainy season based on Weipah distribution according to claim 1, characterized in that, The method obtains the attenuation-free baseline irradiance based on clear-sky theory, and comprehensively considers factors affecting irradiance. Based on the rainfall intensity time-series matrix, it obtains the irradiance time-series matrix, including: Establish a theoretical irradiance calculation model for clear sky and obtain the reference irradiance without attenuation; Taking into account all factors affecting irradiance, a comprehensive irradiance attenuation model coupling multiple influencing factors is established; Based on the rainfall intensity time series matrix, the irradiance time series matrix is ​​obtained by transformation through a comprehensive irradiance attenuation model.

5. The method for constructing photovoltaic power output scenarios during the rainy season based on Weiper distribution according to claim 4, characterized in that, The comprehensive irradiance attenuation model, which takes into account various factors affecting irradiance, and establishes a coupled model of multiple influencing factors, includes: Based on rainfall attenuation, topographic spatial differences, and cloud cover effect, the corresponding rainfall attenuation coefficient, topographic correction coefficient, and cloud cover attenuation coefficient are modeled and obtained respectively. A comprehensive irradiance attenuation model coupling multiple influencing factors is established based on rainfall attenuation coefficient, terrain correction coefficient, and cloud cover attenuation coefficient.

6. The method for constructing photovoltaic power output scenarios during the rainy season based on Weiper distribution according to claim 5, characterized in that, When obtaining the corresponding rainfall attenuation coefficient, terrain correction coefficient, and cloud cover attenuation coefficient, each is modeled independently and a corresponding physical boundary verification mechanism is set.

7. The method for constructing photovoltaic power output scenarios during the rainy season based on Weiper distribution according to claim 1, characterized in that, The photovoltaic power output probability scenario set obtained based on the irradiance time series matrix and combined with ambient temperature conditions includes: Based on historical meteorological data, a temperature time series generation model driven by rainfall intensity is established to generate a temperature time series matrix corresponding to the rainfall intensity time series matrix; A conversion model from actual irradiance and ambient temperature to photovoltaic output is established, and a photovoltaic output time series matrix is ​​generated based on the irradiance time series matrix and the temperature time series matrix. Based on the photovoltaic power output time series matrix and the joint probability density of daily rainfall under the Weibull distribution, the probability of photovoltaic power output scenarios is calculated to obtain the photovoltaic power output probability scenario set during the plum rain season.

8. The method for constructing photovoltaic power output scenarios during the rainy season based on Weiper distribution according to claim 7, characterized in that, Based on historical meteorological data, a temperature time series generation model driven by rainfall intensity is established to generate a temperature time series matrix corresponding to the rainfall intensity time series matrix, including: Based on historical meteorological data, rainfall intensity ranges were divided and temperature parameters for each range were extracted. Based on the rainfall intensity range and the range temperature parameters, a conditional autoregressive model is established and the global baseline temperature parameters are extracted. Based on the rainfall intensity time series matrix and global reference temperature parameters, the temperature at each time moment is obtained and a temperature time series matrix corresponding to the rainfall intensity time series matrix is ​​generated.

9. The method for constructing photovoltaic power output scenarios during the rainy season based on Weipah distribution according to claim 1, characterized in that, Also includes: The timing consistency check and resampling supplement are performed on the photovoltaic output probability scenario set during the plum rain season to obtain the optimal photovoltaic output probability scenario set during the plum rain season. Based on the optimal photovoltaic output probability scenario set during the plum rain season, the statistical characteristics of photovoltaic output at each time step are calculated. Based on the time-by-time statistical characteristics of photovoltaic power output, we identify extreme low-output scenarios and calculate their total probability and expected duration to obtain a statistical feature set of photovoltaic power output during the rainy season.

10. The method for constructing photovoltaic power output scenarios during the rainy season based on Weiper distribution according to claim 9, characterized in that, The process of performing time-series consistency verification and resampling supplementation on the photovoltaic power output probability scenario set during the rainy season specifically involves: constructing time-series consistency constraints and performing group-by-group verification, screening, and resampling supplementation on the photovoltaic power output probability scenario set during the rainy season; wherein, the time-series consistency constraints include constraints on the rate of change of power output at adjacent times and constraints on the smoothness of daily power output. The time-by-time photovoltaic power output statistical characteristics include: time-by-time power output expectation, time-by-time power output standard deviation, and time-by-time power output quantile.

Images