A micro-meteorological station-based distributed photovoltaic power prediction method and related device
By performing gradient analysis on the spatial distribution of irradiance at micro-weather stations, identifying isolated light spot crossing events and dynamically adjusting their weights, the problem of abrupt spatial changes in irradiance in photovoltaic power prediction was solved, resulting in more accurate photovoltaic power prediction and avoiding misjudgments in grid dispatch and resource waste.
Patent Information
- Application Number
- CN202611261190.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-19
- Publication Date
- 2026-11-10
AI Technical Summary
In existing distributed photovoltaic power prediction technologies, the weighting of micro-meteorological irradiance data fusion is fixed, which cannot identify spatial abrupt changes in irradiance caused by isolated local light spots. This results in excessive deviations between photovoltaic power prediction and actual output, leading to problems such as misjudgment in grid dispatch and false alarms from equipment.
By performing spatial gradient analysis on the spatial distribution of irradiance between adjacent micro-weather stations, isolated light spot crossing events are identified, the weights of target stations are dynamically adjusted, and the shadow coverage area is determined by combining the direction and speed of light spot movement. The irradiance sequence is then reconstructed to achieve dynamic adaptive matching.
It effectively improves the accuracy of photovoltaic power prediction, avoids grid dispatch misjudgment and waste of reserve resources, and improves the consistency of prediction results with actual physical processes.
Smart Images

Figure CN123073719A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of distributed photovoltaic power prediction technology, specifically relating to a distributed photovoltaic power prediction method and related device based on a micro-weather station. Background Technology
[0002] Distributed photovoltaic (PV) power generation is a crucial component of the power system's renewable energy expansion. Its output exhibits significant randomness and volatility, directly impacting grid stability and dispatch. PV power forecasting can effectively mitigate output fluctuations and optimize reserve capacity allocation, making it a key technology for ensuring grid supply and demand balance. Currently, PV power forecasting generally relies on collecting irradiance data from local micro-meteorological stations. By fusing observation data from multiple stations, the solar radiation status of the PV array is reconstructed, and the accuracy of this data fusion directly affects the overall effectiveness of the power forecast.
[0003] Currently, in engineering projects, the inverse distance weighted algorithm is mostly used to complete the fusion of micro-meteorological irradiance data. This algorithm only assigns weights based on the spatial distance between the micro-meteorological station and the array center, and obtains the global irradiance through weighted averaging. Under clear, windless conditions and with a uniform irradiance field distribution, the differences in observation data from different stations are small, and this algorithm can meet the needs of conventional forecasting. Existing improved forecasting technologies mainly optimize forecast accuracy for scenarios such as large-scale cloud changes, multi-source meteorological data adaptation, and the coordinated correlation between meteorology and power output, forming a mature basic forecasting scheme.
[0004] Existing solutions cannot identify spatial anomalies in irradiance caused by small-scale isolated light spots, resulting in issues such as fixed weights and distorted fusion. Photovoltaic fields cover large areas, and micro-meteorological stations are widely spaced. Local convection easily forms small-scale isolated cumulus cloud light spots, causing non-uniform irradiance distribution with localized strong light and large-area shadows on the photovoltaic array. Traditional inverse distance-weighted algorithms use fixed spatial weights, which cannot adapt to abrupt changes in irradiance spatial distribution. Even if the data from local stations illuminated by light spots is not representative of the entire region, they will still receive higher weights due to their distance advantage, leading to artificially inflated fused irradiance. Existing improvement technologies have also failed to specifically address this type of local micro-meteorological disturbance, ultimately resulting in excessive deviations between power prediction and actual output, causing numerous problems such as misjudgments in grid dispatching, false alarms from equipment, and waste of reserve resources. Summary of the Invention
[0005] Therefore, it is necessary to provide a distributed photovoltaic power prediction method and related device based on micro-weather stations to address the above-mentioned technical problems, so as to solve the problems in existing distributed photovoltaic power prediction technologies, such as fixed weights for micro-weather irradiance data fusion and the inability to identify spatial abrupt changes in irradiance caused by isolated local light spots.
[0006] In a first aspect, the present invention provides a method for predicting distributed photovoltaic power based on micro-weather stations, comprising the following steps:
[0007] Spatial gradient analysis was performed on the spatial distribution of irradiance between adjacent micro-weather stations to obtain a sequence of spatial gradient analysis results.
[0008] If the spatial gradient analysis result sequence indicates a spatial abrupt change in irradiance, then an isolated spot crossing event is determined to have occurred, and the target micro-weather station where the isolated spot crossing event occurred is obtained.
[0009] Obtain the initial near-station weights of the target micro-weather station, and compress the initial near-station weights to obtain the compressed weights of the target station;
[0010] Based on the direction and speed of movement of isolated light spots, the shadow micro-weather stations within the array shadow coverage area are identified, and the shadow irradiance sequence of the shadow micro-weather stations is obtained.
[0011] Based on the target station compression weight and the shadow irradiance sequence, the irradiance of the target array region is reconstructed to obtain the fused irradiance sequence;
[0012] Photovoltaic power prediction was performed based on the fused irradiance sequence, and the prediction results were obtained.
[0013] Furthermore, spatial gradient analysis was performed on the spatial distribution of irradiance between adjacent micro-weather stations, yielding a sequence of spatial gradient analysis results, including:
[0014] Real-time irradiance sequences and spatial coordinates of multiple micro-weather stations were obtained.
[0015] Determine adjacent micro-weather stations based on spatial coordinates;
[0016] The irradiance difference is calculated based on the real-time irradiance sequence of adjacent micro-weather stations. The spatial distance and directional angle are calculated based on the spatial coordinates of adjacent micro-weather stations. The gradient vector is obtained based on the irradiance difference, spatial distance, and directional angle.
[0017] Feature extraction is performed on the gradient vector to obtain spatial distribution features. These spatial distribution features are then combined according to the time dimension to obtain a sequence of spatial gradient analysis results.
[0018] Furthermore, if the spatial gradient analysis result sequence indicates a spatial abrupt change in irradiance, an isolated spot crossing event is determined to have occurred, and the target micro-weather station where the isolated spot crossing event occurred is identified, including:
[0019] Identify spatial abrupt changes in irradiance within the spatial gradient analysis result sequence, and designate the micro-weather stations corresponding to these spatial abrupt changes as abrupt change nodes.
[0020] The occlusion contour of the isolated light spot is constructed based on the coordinates of the mutation node. If the movement trajectory of the occlusion contour in the continuous time series crosses the preset micro-weather station monitoring area, it is determined that an isolated light spot crossing event has occurred in the preset micro-weather station monitoring area.
[0021] If the duration of an isolated spot crossing event exceeds a preset time threshold, a spot crossing status identifier is generated, and the target micro-weather station is determined from the mutation nodes based on the spot crossing status identifier.
[0022] Furthermore, the initial near-station weights of the target micro-weather station are obtained, and these initial near-station weights are compressed to obtain the compressed weights of the target station, including:
[0023] The spatial distance between the target micro-weather station and its neighboring micro-weather stations is taken as the near-station distance. The reciprocal of the near-station distance is calculated to obtain the distance attenuation base. The initial near-station weight is calculated based on the distance attenuation base.
[0024] If the initial near-station weight is greater than the preset weight threshold, the historical occlusion frequency of the target micro-weather station is obtained as the attenuation factor, and the attenuation factor is multiplied by the initial near-station weight to obtain the weight attenuation amount.
[0025] A nonlinear mapping function is constructed based on the weight decay. The initial near-station weights are then mapped using the nonlinear mapping function to obtain the dynamic compression coefficient. The initial near-station weights are multiplied by the dynamic compression coefficient to obtain the target station compression weight.
[0026] Furthermore, the direction and speed of movement of the isolated light spot are determined based on the time difference of irradiance changes. The determination process includes:
[0027] Extract the trough times of the real-time irradiance sequences of the target micro-weather station and adjacent micro-weather stations, and determine the time delay based on the temporal difference between the trough times of the target micro-weather station and the trough times of adjacent micro-weather stations.
[0028] The distance vector is calculated based on the spatial coordinates of the target micro-weather station and adjacent micro-weather stations. The velocity scalar of the isolated spot is calculated based on the distance vector and the time delay. The orientation angle of the isolated spot is calculated based on the distance vector.
[0029] The direction and speed of movement of the isolated light spot are determined based on the velocity scalar and the direction angle.
[0030] Furthermore, based on the direction and speed of movement of isolated light spots, the shadow micro-weather stations within the array's shadow coverage area are identified, and the shadow irradiance sequences of the shadow micro-weather stations are obtained, including:
[0031] Based on the direction and speed of movement of the isolated light spot, combined with the initial outline of the isolated light spot, the projection trajectory of the isolated light spot in the target array region is determined.
[0032] Determine whether the spatial coordinates of the candidate micro-weather station are within the projection trajectory. If so, identify the corresponding candidate micro-weather station as a shadow micro-weather station.
[0033] Based on the movement speed of the isolated light spot and the spatial coordinates of the shadow micro-weather station, the coverage time period of the isolated light spot over the shadow micro-weather station is calculated. The real-time irradiance sequence of the shadow micro-weather station is then extracted based on the coverage time period to obtain the shadow irradiance sequence.
[0034] Furthermore, based on the target station compression weights and the shadow irradiance sequence, the irradiance of the target array region is reconstructed to obtain a fused irradiance sequence, including:
[0035] The initial station weights of the target micro-weather stations are updated based on the target station compression weights. Based on the updated micro-weather station weights and the shadow irradiance sequence, spatial interpolation is performed on the target array region to obtain the reconstructed input sequence.
[0036] The coefficient of variation and skewness coefficient are calculated based on the reconstructed input sequence. If the coefficient of variation is greater than a preset threshold, statistical features are constructed based on the coefficient of variation and skewness coefficient.
[0037] By fusing statistical features with the reconstructed input sequence, a fused irradiance sequence is obtained.
[0038] Secondly, the present invention provides a distributed photovoltaic power prediction device based on a micro-weather station, comprising:
[0039] The irradiance gradient analysis module is used to perform spatial gradient analysis on the spatial distribution of irradiance between adjacent micro-weather stations and obtain a sequence of spatial gradient analysis results.
[0040] The target micro-weather station acquisition module is used to determine the occurrence of an isolated light spot crossing event if the spatial gradient analysis result sequence indicates that there is a spatial abrupt change in irradiance, and to obtain the target micro-weather station where the isolated light spot crossing event occurred.
[0041] The weight compression module is used to obtain the initial near-station weights of the target micro-weather station and compress the initial near-station weights to obtain the compressed weights of the target station.
[0042] The shadow irradiance sequence acquisition module is used to determine the shadow micro-weather stations within the array shadow coverage area based on the movement direction and speed of isolated light spots, and to obtain the shadow irradiance sequence of the shadow micro-weather stations.
[0043] The fused irradiance sequence generation module is used to reconstruct the irradiance of the target array region based on the target station compression weight and the shadow irradiance sequence to obtain the fused irradiance sequence.
[0044] The photovoltaic power prediction module is used to predict photovoltaic power based on the fused irradiance sequence and obtain the prediction results.
[0045] Thirdly, the present invention provides a computer device, the device including a processor and a memory:
[0046] The memory is used to store computer programs and send the instructions of the computer programs to the processor;
[0047] The processor executes the following steps according to the instructions of the computer program:
[0048] Spatial gradient analysis was performed on the spatial distribution of irradiance between adjacent micro-weather stations to obtain a sequence of spatial gradient analysis results.
[0049] If the spatial gradient analysis result sequence indicates a spatial abrupt change in irradiance, then an isolated spot crossing event is determined to have occurred, and the target micro-weather station where the isolated spot crossing event occurred is obtained.
[0050] Obtain the initial near-station weights of the target micro-weather station, and compress the initial near-station weights to obtain the compressed weights of the target station;
[0051] Based on the direction and speed of movement of isolated light spots, the shadow micro-weather stations within the array shadow coverage area are identified, and the shadow irradiance sequence of the shadow micro-weather stations is obtained.
[0052] Based on the target station compression weight and the shadow irradiance sequence, the irradiance of the target array region is reconstructed to obtain the fused irradiance sequence;
[0053] Photovoltaic power prediction was performed based on the fused irradiance sequence, and the prediction results were obtained.
[0054] Fourthly, the present invention provides a computer-readable storage medium on which a computer program is stored, and when executed by a processor, the computer program performs the following steps:
[0055] Spatial gradient analysis was performed on the spatial distribution of irradiance between adjacent micro-weather stations to obtain a sequence of spatial gradient analysis results.
[0056] If the spatial gradient analysis result sequence indicates a spatial abrupt change in irradiance, then an isolated spot crossing event is determined to have occurred, and the target micro-weather station where the isolated spot crossing event occurred is obtained.
[0057] Obtain the initial near-station weights of the target micro-weather station, and compress the initial near-station weights to obtain the compressed weights of the target station;
[0058] Based on the direction and speed of movement of isolated light spots, the shadow micro-weather stations within the array shadow coverage area are identified, and the shadow irradiance sequence of the shadow micro-weather stations is obtained.
[0059] Based on the target station compression weight and the shadow irradiance sequence, the irradiance of the target array region is reconstructed to obtain the fused irradiance sequence;
[0060] Photovoltaic power prediction was performed based on the fused irradiance sequence, and the prediction results were obtained.
[0061] In summary, this invention provides a distributed photovoltaic power prediction method and related device based on micro-weather stations. The method first identifies abrupt changes and determines isolated light spot crossing events and target stations by performing spatial gradient analysis of irradiance between adjacent micro-weather stations. Then, it compresses the initial near-station weights of the target station, solving the problem of weight solidification and artificially high weights caused by distance advantages of local stations. Simultaneously, it accurately locates the shadowed micro-weather stations in the shadowed area based on the direction and speed of light spot movement and obtains their irradiance sequences. Finally, it reconstructs the irradiance of the target array area by combining the compressed weights of the target station with the shadowed irradiance sequences, achieving dynamic adaptive matching for non-uniform abrupt irradiance spatial distribution. This completely overcomes the fusion distortion problem under micro-weather disturbances, ultimately effectively improving the accuracy of photovoltaic power prediction and avoiding grid dispatch misjudgments, equipment false alarms, and waste of reserve resources caused by excessive deviations between prediction and actual output. Attached Figure Description
[0062] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0063] Figure 1 This is a flowchart illustrating a distributed photovoltaic power prediction method based on a micro-weather station, as described in one embodiment of the present invention.
[0064] Figure 2 This is a block diagram of a distributed photovoltaic power prediction device based on a micro-weather station, as described in one embodiment of the present invention.
[0065] Figure 3 This is a block diagram of a computer device according to one embodiment of the present invention. Detailed Implementation
[0066] To make the objectives, features, and advantages of this invention more apparent and understandable, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.
[0067] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0068] It should be noted that the user information (including but not limited to user images, user portrait information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of related data must comply with relevant regulations.
[0069] In one embodiment, such as Figure 1 As shown, a distributed photovoltaic power prediction method based on micro-weather stations is provided, including the following steps:
[0070] S101: Spatial gradient analysis of the spatial distribution of irradiance between adjacent micro-weather stations is performed to obtain a sequence of spatial gradient analysis results.
[0071] Among them, a micro-weather station refers to a station deployed in a photovoltaic field for real-time monitoring of local meteorological data. The real-time irradiance sequence is a continuous solar radiation data stream arranged by timestamps. The spatial coordinates are the precise three-dimensional geographical location information of the micro-weather station. The instantaneous irradiance value is the irradiance reading at a specific moment. Spatial gradient analysis is a method to quantify the degree of drastic change in irradiance over spatial physical distance. The spatial gradient analysis result sequence is a dynamic sequence composed of spatial gradient values on multiple consecutive time sections spliced in chronological order.
[0072] Optionally, real-time irradiance sequences and spatial coordinates of multiple micro-weather stations can be obtained, instantaneous irradiance values can be extracted, adjacent micro-weather stations can be determined based on spatial coordinates, and spatial gradient analysis of the spatial distribution of irradiance between adjacent micro-weather stations can be performed based on instantaneous irradiance values and spatial coordinates to obtain a spatial gradient analysis result sequence.
[0073] S102: If the spatial gradient analysis result sequence indicates a spatial abrupt change in irradiance, then an isolated spot crossing event is determined to have occurred, and the target micro-weather station where the isolated spot crossing event occurred is obtained.
[0074] Among them, spatial abrupt change in irradiance refers to the phenomenon of a cliff-like change in solar radiation intensity within a very short physical distance, isolated spot crossing event is a meteorological phenomenon in which local strong light transmitted through cloud gaps sweeps rapidly across the ground, and the target micro-weather station is the micro-weather station most significantly affected by spot irradiation.
[0075] Optionally, the current gradient value in the spatial gradient analysis result sequence is read. If the current gradient value indicates a spatial abrupt change in irradiance, an isolated spot crossing event is determined, a spot crossing status identifier is generated, and the target micro-weather station where the isolated spot crossing event occurred is determined based on the spot crossing status identifier.
[0076] S103: Obtain the initial near-station weights of the target micro-weather station, and compress the initial near-station weights to obtain the compressed weights of the target station.
[0077] Among them, the initial near-station weight is an indicator that quantifies the degree of direct physical influence on the target micro-meteorological station in the light spot crossing event. The compression processing is a numerical mapping mechanism that applies nonlinear suppression to extremely high weights that exceed the conventional physical limits. The target station compressed weight is weight data with good numerical stability after dynamic convergence.
[0078] Optionally, based on the light spot crossing status identifier and the target micro-weather station, the initial near-station weight of the target micro-weather station is obtained, and the initial near-station weight is dynamically compressed to obtain the updated target station compressed weight.
[0079] S104: Based on the direction and speed of movement of isolated light spots, determine the shadow micro-weather stations within the array shadow coverage area and obtain the shadow irradiance sequence of the shadow micro-weather stations.
[0080] Among them, the direction of motion and the speed of motion are vector parameters of the light spot moving in space, the array shadow coverage area is the projection band formed by the light spot's movement trajectory on the photovoltaic array, the shadow micro-weather station is the station located within the projection band that will be swept by the shadow of the light spot, and the shadow irradiance sequence is the irradiance data stream recorded by the shadow micro-weather station that includes the shading process.
[0081] Optionally, the temporal differences in irradiance changes can be analyzed by using the real-time irradiance sequence of the target micro-weather station and the real-time irradiance sequence of adjacent micro-weather stations to determine the movement direction and speed of isolated light spots; based on the movement direction and speed of isolated light spots, the shadow micro-weather stations within the array shadow coverage area can be identified, and the shadow irradiance sequence of the shadow micro-weather stations can be obtained.
[0082] S105: Based on the target station compression weight and the shadow irradiance sequence, the irradiance of the target array region is reconstructed to obtain the fused irradiance sequence.
[0083] The target array region is the physical area where the photovoltaic array is located. The reconstruction process is the process of converting the point monitoring data of discrete sites into continuous area data and extracting features. The fused irradiance sequence is a compact feature vector that includes the absolute shading intensity and the spatial light gradient variation law.
[0084] Optionally, the weights of the target micro-weather stations are updated according to the target station compression weights, and the target array region is spatially interpolated and reconstructed according to the updated weights of each micro-weather station and the shadow irradiance sequence to obtain the reconstruction input sequence. The irradiance statistical feature values are extracted according to the reconstruction input sequence to obtain the fused irradiance sequence.
[0085] S106: Photovoltaic power prediction based on fused irradiance sequence, and the prediction results are obtained.
[0086] Photovoltaic power prediction is a process of extrapolating future photovoltaic power generation using meteorological data, and the prediction result is the power estimate output by the model.
[0087] Optionally, the fused irradiance sequence can be used as the main feature vector to input into the distributed photovoltaic power prediction model to complete the prediction input construction and output the photovoltaic power prediction results.
[0088] The method provided in this embodiment accurately identifies isolated light spot crossing events and target sites through spatial gradient analysis, dynamically compresses the weights of target sites to break the weight solidification and artificial inflated weights caused by distance advantages, tracks the movement trajectory of light spots to predict the shadow coverage area and extracts the shadow irradiance sequence, and then combines the compressed weights and shadow sequence to reconstruct the irradiance of the target array area. This achieves dynamic adaptive matching for non-uniform and abrupt irradiance spatial distribution, effectively overcomes the fusion distortion problem under micro-meteorological disturbances, provides high-fidelity front-end data support for photovoltaic power prediction, and significantly improves prediction accuracy.
[0089] In an exemplary embodiment, spatial gradient analysis is performed on the spatial distribution of irradiance between adjacent micro-weather stations to obtain a sequence of spatial gradient analysis results, including:
[0090] S201: Obtain the real-time irradiance sequence of multiple micro-weather stations and the spatial coordinates of multiple micro-weather stations.
[0091] Among them, the real-time irradiance sequence is a set of irradiance values strictly aligned with timestamps collected by the high-frequency sensor of the micro-weather station within a continuous time window, and the spatial coordinates are precise three-dimensional geographical location information including longitude, latitude and altitude obtained through the Global Positioning System.
[0092] Specifically, within the deployment area of large-scale photovoltaic power plants or distributed photovoltaic arrays, multiple micro-weather stations are typically installed in a grid pattern to monitor local weather changes in real time. These micro-weather stations continuously collect solar radiation data through sensors, forming a real-time irradiance sequence arranged by timestamps. Simultaneously, each micro-weather station obtains precise three-dimensional spatial coordinates, including longitude, latitude, and altitude, through the Global Positioning System during installation. This coordinate data provides a basic geographical location reference for subsequent spatial analysis.
[0093] S202: Determine adjacent micro-weather stations based on spatial coordinates.
[0094] Among them, adjacent micro-meteorological stations are station pairs that have a direct proximity relationship in spatial topology.
[0095] In one possible implementation, the system uses a spatial topology algorithm to construct an adjacency network between multiple micro-weather stations based on their acquired spatial coordinates. Specifically, this can be achieved by constructing a Delaunay triangulation network, treating each micro-weather station as a node in a planar or three-dimensional space, and forming a series of non-intersecting triangles by connecting adjacent nodes. Within these triangles, two micro-weather stations connected by the same edge are considered adjacent. This method objectively reflects the spatial proximity between stations, avoiding the problems of isolated stations or overly dense connections that may occur when relying solely on fixed distance thresholds to define adjacency relationships.
[0096] S203: Calculate the irradiance difference based on the real-time irradiance sequence of adjacent micro-weather stations, calculate the spatial distance and direction angle based on the spatial coordinates of adjacent micro-weather stations, and obtain the gradient vector based on the irradiance difference, spatial distance, and direction angle.
[0097] Among them, the irradiance difference is the absolute difference in instantaneous irradiance between two adjacent micro-meteorological stations at the same moment, the direction angle is the relative azimuth angle of the spatial line connecting the two adjacent micro-meteorological stations, and the gradient vector is a composite physical quantity that includes the amplitude and direction of irradiance change.
[0098] It should be noted that spatial gradient analysis of the irradiance distribution between adjacent micro-weather stations is a key aspect of this invention. Spatial gradient analysis essentially quantifies the drastic change in irradiance over spatial physical distance. In photovoltaic power generation scenarios, due to factors such as cloud movement, terrain shading, or localized haze, solar irradiance often differs between adjacent areas. By calculating the spatial gradient, the magnitude and direction of this difference can be accurately captured, thereby identifying the boundaries of abrupt changes in irradiance. This is of significant indicative importance for assessing the impact of localized shading on the overall power generation efficiency of photovoltaic arrays.
[0099] Specifically, the spatial gradient analysis involves a comprehensive calculation of the irradiance difference and spatial distance between adjacent micro-weather stations. First, the system calculates the actual physical distance (spatial distance) between two adjacent micro-weather stations using the spherical distance formula or the three-dimensional Euclidean distance formula, based on their spatial coordinates. Then, the system extracts the instantaneous irradiance values of these two adjacent micro-weather stations at the same moment and calculates the absolute difference in irradiance between them. Next, dividing this absolute difference in irradiance by the actual physical distance between them yields the instantaneous spatial gradient value between the two adjacent micro-weather stations. This gradient value represents the change in irradiance per unit distance; a larger value indicates more drastic irradiance changes in the local area, possibly indicating the area is in a shadow transition zone at the edge of a cloud layer; a smaller value indicates a relatively uniform irradiance distribution in the area. By traversing all pairs of adjacent micro-weather stations, the system can calculate the spatial gradient values along various local connecting lines within the entire monitoring area.
[0100] In another implementation, considering the impact of complex terrain on irradiance, spatial gradient analysis can also incorporate an elevation correction factor. For example, when the altitude difference between two adjacent micro-meteorological stations is significant, the system can adjust the physical distance based on the elevation difference when calculating the spatial gradient, or perform benchmark normalization on the instantaneous irradiance values based on the variation of atmospheric transmittance with altitude. This effectively eliminates static interference from terrain undulations, resulting in a more pure spatial gradient analysis sequence that reflects the spatial distribution characteristics of irradiance caused by dynamic changes in meteorological conditions.
[0101] S204: Extract spatial distribution features from the gradient vector, and combine the spatial distribution features according to the time dimension to obtain a sequence of spatial gradient analysis results.
[0102] Among them, spatial distribution features are pattern features that reflect spatial differences extracted from gradient sequences. The spatial gradient analysis result sequence is a dynamic sequence of spatial distribution features from multiple consecutive time sections spliced together in the order of temporal evolution.
[0103] Preferably, after the above-described traversal calculations, the system obtains a sequence of spatial gradient analysis results between adjacent micro-weather stations. This sequence not only includes the identification information of each pair of adjacent micro-weather stations but also records their spatial gradient values across multiple consecutive time sections. As time progresses, the system repeatedly extracts instantaneous irradiance values and performs spatial gradient calculations, thereby piecing together discrete instantaneous gradient values into a dynamic sequence in chronological order. This sequence comprehensively depicts the dynamic process of the spatial difference in irradiance within the photovoltaic power station area over time, and can intuitively reflect the spatial movement trajectory and diffusion trend of obstructions such as clouds.
[0104] This embodiment constructs a spatial topology network of micro-weather stations and introduces terrain correction for spatial gradient calculation, which can objectively and accurately extract the analysis result sequence reflecting the spatial difference evolution process of irradiance in the photovoltaic array area.
[0105] In an exemplary embodiment, if the spatial gradient analysis result sequence indicates a spatial abrupt change in irradiance, an isolated spot crossing event is determined to have occurred, and the target micro-weather station where the isolated spot crossing event occurred is obtained, including:
[0106] S301: Identify spatial abrupt changes in irradiance in the spatial gradient analysis result sequence, and take the micro-weather stations corresponding to the spatial abrupt changes in irradiance as abrupt change nodes.
[0107] Among them, the spatial mutation of irradiance is an abnormal extreme point where the absolute value of the current gradient value exceeds the preset benchmark threshold, and the mutation node is the micro-weather station node that participates in constituting the mutation feature.
[0108] Specifically, after acquiring the spatial gradient analysis result sequence between adjacent micro-weather stations, the system reads the current gradient value in the sequence one by one to determine whether there is a spatial abrupt change in irradiance within the monitoring area. This step aims to filter out abnormal extreme points from the continuous spatial rate of change. Determining whether the current gradient value indicates a spatial abrupt change in irradiance requires a rigorous comparison mechanism. The system typically pre-sets a spatial gradient benchmark threshold, which can be a fixed empirical value or a dynamic benchmark calculated based on the gradient distribution under historical clear weather conditions. When the absolute value of the current gradient value exceeds this benchmark threshold, the system identifies a spatial abrupt change in irradiance between the corresponding two adjacent micro-weather stations. This abrupt change means that within a very short physical distance, the solar radiation intensity has experienced a precipitous increase or decrease, usually corresponding to the shadow boundary at the cloud edge. To improve the reliability of the determination, the system also performs a secondary verification by combining the average background irradiance before and after the abrupt change to eliminate false abrupt changes caused by momentary sensor malfunctions.
[0109] S302: Construct the occlusion contour of the isolated light spot based on the coordinates of the mutation node. If the movement trajectory of the occlusion contour in the continuous time series crosses the preset micro-weather station monitoring area, it is determined that an isolated light spot crossing event has occurred in the preset micro-weather station monitoring area.
[0110] Among them, the occlusion contour is a closed polygonal region with a high gradient boundary formed by multiple mutation nodes, the movement trajectory is the spatial displacement path of the closed boundary within a continuous time section, and the isolated spot crossing event is a meteorological event in which a closed region with high irradiance undergoes a clear directional translation and does not dissipate over a large area.
[0111] It should be noted that isolated light spot crossing events are a typical micrometeorological phenomenon in photovoltaic power generation scenarios that significantly impacts the local array's operational status. In cloudy or overcast conditions, thick cloud cover typically blocks most sunlight, resulting in low overall irradiance. However, when small gaps or holes exist within the clouds, direct sunlight can penetrate these gaps, projecting a bright area of limited size but extremely high irradiance onto the ground—the isolated light spot. As high-altitude winds shift, these cloud gaps move, causing the isolated light spot to rapidly sweep across the ground where the photovoltaic array is located, forming a crossing event. The system's logic for determining the occurrence of this event lies in identifying specific topological patterns within the spatial gradient sequence.
[0112] In practice, after detecting a localized abrupt change in positive irradiance, the system immediately retrieves gradient data from other adjacent micro-weather stations in the vicinity. If a closed polygonal region enclosed by multiple micro-weather stations is found, and the spatial gradient values on its boundaries all exhibit abrupt changes with a sharp increase from the outside in, and the absolute value of irradiance inside this closed region is much higher than the external background value, the system initially identifies an isolated spot. Subsequently, the system combines continuous cross-sectional data in the time dimension to track the spatial displacement of this high-gradient closed boundary. If this closed boundary exhibits translational characteristics with a clear direction and velocity in multiple consecutive time sections, and no large-scale dissipation or merging occurs, the system ultimately determines that a complete isolated spot crossing event has occurred.
[0113] In another possible implementation, to address interference from complex cloud formations, the system introduces morphological constraints when identifying isolated light spot crossing events. For example, the system calculates the equivalent diameter and area of the light spot region enclosed by a high-gradient boundary. Only when this area falls within a preset range and its shape remains relatively stable is it identified as an isolated light spot. If the area is too large, it may be at the edge of a large clear-sky area; if the area is too small, it may be noise caused by local reflections. This morphological filtering ensures that the identified crossing events have genuine physical meaning.
[0114] S303: If the duration of an isolated spot crossing event is greater than a preset time threshold, a spot crossing status identifier is generated, and the target micro-weather station is determined from the mutation nodes based on the spot crossing status identifier.
[0115] Among them, the duration is the length of time from the occurrence to the end of the spot crossing event, the spot crossing status identifier is a unique digital label that encapsulates the event attribute information, and the target micro-weather station is one of the stations most significantly affected by the spot irradiation.
[0116] For example, after successfully determining that an isolated light spot crossing event has occurred, the system immediately generates and obtains a light spot crossing status identifier. This status identifier is a structured data packet encapsulating rich event attribute information. Specifically, it includes the precise timestamp of the event trigger, the direction of the light spot's movement vector, the estimated crossing duration, and the event's severity rating. This identifier is equivalent to attaching a unique digital label to the meteorological event, enabling the system to transform the originally disordered low-level meteorological data into a structured event record. Simultaneously, the system also needs to obtain the target micro-meteorological stations where the isolated light spot crossing event occurred. Specifically, the system traverses all micro-meteorological station nodes that participate in forming the high gradient boundary of the isolated light spot and those located within the light spot. Based on the peak irradiance recorded by each station during the crossing event and the duration of high irradiance, the system selects several stations most significantly affected by the light spot irradiation and marks them as target micro-meteorological stations. The precise spatial coordinates of these target micro-meteorological stations are extracted to depict the actual sweeping trajectory of the light spot on the ground. By accurately locating these target micro-meteorological stations, the system can accurately determine the specific geographical range of localized intense light irradiation.
[0117] This embodiment accurately identifies isolated light spot crossing events from both spatial topological patterns and temporal displacement trajectories by setting gradient thresholds and morphological constraints, and generates structured identifiers and target site information, effectively eliminating interference from pseudo-mutations.
[0118] In an exemplary embodiment, the initial near-station weights of the target micro-weather station are obtained, and the initial near-station weights are compressed to obtain the compressed weights of the target station, including:
[0119] S401: The spatial distance between the target micro-weather station and the adjacent micro-weather station is taken as the near-station distance. The reciprocal of the near-station distance is calculated to obtain the distance attenuation base. The initial near-station weight is calculated based on the distance attenuation base.
[0120] Among them, the nearest station distance is the actual physical distance between the target micro-weather station and the adjacent station, the distance attenuation base is the reciprocal of the distance, and the initial nearest station weight is an indicator that comprehensively quantifies the degree of direct physical influence of the light spot on the target station by the spatial and temporal dimensions.
[0121] Specifically, after successfully acquiring the light spot crossing status identifier and the target micro-weather station where an isolated light spot crossing event occurred, the system further initiates a spatial impact assessment process for these target stations. Obtaining the initial near-station weights for the target micro-weather stations requires comprehensive consideration of the dynamic interaction between spatial and temporal dimensions. The system analyzes the light spot movement vector direction and estimated crossing duration encapsulated in the light spot crossing status identifier, and performs joint calculations based on the precise spatial coordinates of each target micro-weather station. In the spatial dimension, the system constructs a distance attenuation model with the instantaneous center of the light spot as the origin; the closer the target micro-weather station is to the center of the light spot, the higher its assigned spatial base weight. In the temporal dimension, the system calculates the actual cumulative duration of each target micro-weather station in a high-irradiance state; the longer the duration, the greater the temporal base weight. Subsequently, the system weights and fuses the spatial and temporal base weights to calculate the initial near-station weights for each target micro-weather station. This two-dimensional calculation method accurately reflects the differentiated irradiance intensity on stations at different geographical locations during the light spot sweeping process.
[0122] S402: If the initial near-station weight is greater than the preset weight threshold, the historical occlusion frequency of the target micro-weather station is obtained as the attenuation factor, and the attenuation factor is multiplied by the initial near-station weight to obtain the weight attenuation amount.
[0123] Among them, the weight threshold is the critical value for determining whether the weight has polarized expansion, the historical occlusion frequency is the frequency statistics of past occlusion events of the target site, the attenuation factor is a dynamic parameter used to adjust the weight overflow part, and the weight attenuation amount is the numerical result of multiplying the attenuation factor by the initial weight.
[0124] It should be noted that in extreme isolated light spot crossing events, due to the lensing effect of cloud gaps, the irradiance at the center of the light spot may reach abnormally high values. This situation causes a sharp expansion of the initial near-station weights acquired by the target micro-meteorological station in the core region of the light spot, leading to severe polarization of the weight distribution throughout the monitoring network. To address this weight polarization problem, the system needs to dynamically compress the initial near-station weights. Dynamic compression is essentially a nonlinear numerical mapping mechanism. Its core logic lies in maintaining linear or near-linear transmission of weight values within a reasonable range, while applying a gradual suppression effect to extremely high weight values that exceed conventional physical limits. Through this processing, the system can effectively converge the absolute numerical range of the weights while preserving the differences in the relative influence between stations.
[0125] S403: Construct a nonlinear mapping function based on the weight decay, perform weight mapping on the initial near-station weights through the nonlinear mapping function to obtain the dynamic compression coefficient, and multiply the initial near-station weights by the dynamic compression coefficient to obtain the target station compression weight.
[0126] Among them, the nonlinear mapping function is a logarithmic or exponential function whose attenuation coefficient gradually decreases as the overflow increases; the dynamic compression coefficient is a specific value used to adjust the weight after mapping; and the target station compression weight is weight data with good numerical stability and anti-interference ability after dynamic convergence.
[0127] For example, during the dynamic compression process, the system pre-sets a weight polarization critical threshold. When the system iterates through the initial near-station weights of all target micro-weather stations, if it finds that the initial near-station weight of a station is lower than or equal to the critical threshold, the system will directly retain that value or only make minor smoothing adjustments. However, when the initial near-station weight of a station exceeds the critical threshold, the system will trigger a nonlinear compression algorithm. In specific operation, the system calculates the overflow portion exceeding the critical threshold and applies a dynamic attenuation coefficient to this overflow portion. This attenuation coefficient is not fixed but gradually decreases as the overflow amount increases, exhibiting a logarithmic or exponential attenuation characteristic. The system re-adds the attenuated overflow portion to the critical threshold to obtain the updated target station compressed weight for that target micro-weather station. This segmented dynamic compression strategy prevents the infinite amplification of the core station weights and avoids the loss of data features caused by direct truncation.
[0128] In one possible implementation, key parameters in the dynamic compression process can also be adaptively adjusted. The system extracts the event severity rating information from the spot crossing status identifier and dynamically adjusts the initial benchmarks of the aforementioned weight polarization critical threshold and dynamic attenuation coefficient based on this rating. For example, when the event severity rating is high, meaning the spot irradiance is extremely strong and moving rapidly, the system will appropriately lower the critical threshold and enhance the attenuation force to more aggressively suppress potential weight distortion. After the above series of rigorous calculation and adjustment steps, the system finally obtains the updated target station compression weights.
[0129] This embodiment calculates the initial near-site weights jointly in both spatial and temporal dimensions, and introduces a nonlinear mapping mechanism to dynamically compress and adaptively adjust the polarization weights caused by the lensing effect of light spots. This effectively prevents the infinite amplification of the core site weights and ensures the balance and representativeness of the entire array of sensing data.
[0130] In an exemplary embodiment, the direction and speed of movement of the isolated light spot are determined based on the time difference of irradiance changes, and the determination process includes:
[0131] S501: Extract the trough times of the real-time irradiance sequences of the target micro-weather station and adjacent micro-weather stations, and determine the time delay based on the temporal difference between the trough times of the target micro-weather station and the trough times of adjacent micro-weather stations.
[0132] Among them, the trough time is the timestamp when the irradiance drops sharply to a minimum value in the real-time irradiance sequence, the time difference is the time difference between different stations sensing the same spot feature, and the time delay is the effective time difference value that is greater than the preset difference threshold.
[0133] Specifically, after locating the target micro-weather station where an isolated light spot crossing event has occurred, the system simultaneously retrieves monitoring data from the target micro-weather station and its neighboring stations. The system continuously receives real-time irradiance sequences collected by these stations via the underlying data bus. This real-time irradiance sequence is not a single instantaneous measurement, but rather a set of strictly timestamped irradiance values collected by the micro-weather station's high-frequency sensors within a continuous time window. To eliminate the influence of sensor noise and high-frequency environmental disturbances, the system performs a moving average filter on the original sequence to extract a low-frequency envelope that accurately reflects the physical processes of light spot occlusion and transmission, thereby constructing a smooth and characteristic standard irradiance sequence.
[0134] Analyzing the temporal differences in irradiance variation is a crucial step in the entire dynamic tracking process of the solar spot. Because the movement of an isolated solar spot in space is continuous, micro-weather stations at different spatial locations will sense drastic fluctuations in irradiance sequentially as the spot sweeps across the geographical area where the photovoltaic array is located. The system employs a cross-correlation analysis algorithm to quantify this sequential order. The system uses the real-time irradiance sequence of the target micro-weather station as the baseline signal and the real-time irradiance sequences of adjacent micro-weather stations as the comparison signal. By progressively shifting the comparison signal along the time axis and continuously calculating the inner product between the two signal sequences, the system can plot the cross-correlation function curve. When the cross-correlation function reaches its maximum peak, the corresponding shift time represents the precise temporal difference in how the two stations perceive the same edge or core feature of the solar spot. This analysis method based on full-sequence morphological matching effectively overcomes the limitation of single-point threshold triggering mechanisms, which are easily affected by uneven local cloud thickness. In actual business operations, the system sets a sliding time window and performs cross-correlation calculations only on the sequence segments within the window to accommodate the slight deformations that may occur in the light spot morphology during long-distance movement, ensuring the rigor of time difference extraction.
[0135] For example, in extreme weather scenarios with extremely sharp light spot edges, the system can also employ a feature point dynamic regularization strategy to assist in the extraction of temporal differences. The system calculates the first derivative of the real-time irradiance sequence to identify abrupt changes in irradiance, such as sharp increases or decreases. By comparing the timestamp differences between the target micro-weather station and adjacent micro-weather stations at similar abrupt change edges, the system can pinpoint the absolute time delay of the light spot traversing different stations with extremely high accuracy. Furthermore, to ensure the effectiveness of the temporal difference analysis, the system follows strict spatial topology rules when selecting adjacent micro-weather stations. Using the target micro-weather station as a benchmark, the system selects at least two non-collinear adjacent micro-weather stations within a preset effective sensing radius.
[0136] S502: Calculate the distance vector based on the spatial coordinates of the target micro-weather station and adjacent micro-weather stations, calculate the velocity scalar of the isolated spot based on the distance vector and time delay, and calculate the direction angle of the isolated spot based on the distance vector.
[0137] Among them, the distance vector is a geometric parameter that includes the straight-line distance between stations and the relative azimuth angle, the velocity scalar is the ratio of distance to time delay, and the direction angle is the azimuth angle of the light spot movement.
[0138] Understandably, after obtaining the precise time difference, the system immediately initiates the calculation process to determine the motion direction and velocity of the isolated light spot. This process essentially maps the time-dimensional delay into a spatial-dimensional physical motion vector. The system extracts the absolute geographic coordinates of the target micro-weather station and each adjacent micro-weather station, calculating the spatial straight-line distance and relative azimuth angle between the stations. Assuming the light spot moves at a uniform linear velocity within a local area, the system uses the spatial distance vector between adjacent micro-weather stations and the target micro-weather station, divided by the corresponding time difference, to initially calculate the apparent velocity component of the light spot in that direction. Subsequently, the system performs vector synthesis of the apparent velocity components provided by multiple adjacent micro-weather stations. By constructing a set of nonlinear equations containing the actual motion velocity of the light spot, the actual motion direction angle, and the relative azimuth angles of each station, the system can inversely solve for the true motion direction and absolute velocity of the isolated light spot at the current moment. In this solution model, the system also introduces a terrain undulation compensation factor. For micro-weather stations deployed in mountainous or hilly areas, the system performs three-dimensional projection correction on the spatial straight-line distance based on the elevation difference, thereby eliminating the geometric distortion caused by the terrain slope to the calculation of the light spot movement speed.
[0139] S503: Determine the motion direction and speed of the isolated light spot based on the velocity scalar and the direction angle.
[0140] The direction of motion and the speed of motion are the actual movement vector parameters of the light spot obtained through vector synthesis and optimization.
[0141] Preferably, in centralized photovoltaic power station scenarios with densely deployed micro-weather stations, the system typically acquires redundant data from adjacent micro-weather stations. In this case, the system introduces a least-squares optimization framework to handle multiple combinations of time differences and spatial coordinates. By constructing an objective error function, the system iteratively adjusts the estimated motion direction and velocity parameters to minimize the sum of squared errors between the theoretically calculated and actual measured time delays of all observation stations, thereby outputting the globally optimal light spot motion vector. In one possible implementation, if the system discovers a significant deviation in the velocity component derived from an adjacent micro-weather station compared to other stations during the calculation process, the system determines that the data from that station is contaminated by local bird shading or transient equipment failure. The system automatically removes the real-time irradiance sequence from this abnormal station and re-performs time difference analysis and vector calculation using the remaining healthy station data to ensure the reliability of the final output motion direction and velocity.
[0142] This embodiment accurately extracts the time delay between multiple stations through cross-correlation analysis and dynamic warping algorithm, and combines spatial coordinates for vector synthesis and least squares optimization solution, effectively overcoming the interference of single-point threshold triggering, and accurately and reliably restoring the true motion vector parameters of isolated light spots.
[0143] In an exemplary embodiment, based on the movement direction and speed of isolated light spots, shadow micro-weather stations within the array's shadow coverage area are determined, and shadow irradiance sequences of the shadow micro-weather stations are obtained, including:
[0144] S601: Based on the direction and speed of movement of the isolated light spot, and combined with the initial outline of the isolated light spot, determine the projection trajectory of the isolated light spot in the target array region.
[0145] Among them, the initial outline of the light spot is the spatial scale and geometric shape of the light spot at the starting point, and the projection trajectory is the band-shaped envelope formed by the light spot extending on the array plane over time.
[0146] Specifically, after obtaining the absolute direction and velocity of the isolated light spot, the system immediately enters the dynamic simulation stage of the array's shadow coverage area. Based on the aforementioned motion vector parameters and combined with the global spatial topology of the photovoltaic array, the system proactively identifies the micro-weather stations that will be swept by the light spot's shadow within a certain period. The core of determining the array's shadow coverage area lies in constructing a dynamic sweeping geometric model of the isolated light spot. The system first needs to determine the initial spatial scale of the light spot. In one possible implementation, the system extracts the duration of the light spot shading perceived by the target micro-weather station and multiplies it by the calculated light spot velocity to estimate the equivalent chord length of the light spot in the direction of motion. Subsequently, starting from the geographical coordinates of the target micro-weather station, the system constructs a two-dimensional polygonal region extending over time along the determined direction of motion; this region is the theoretical sweeping band of the array's shadow coverage area. The system adds a preset lateral expansion tolerance distance on both sides of this sweeping band, forming a strip-shaped envelope with a certain width. The mathematical representation of this banded envelope in the geographic coordinate system consists of a set of boundary coordinate points that are dynamically updated over time.
[0147] S602: Determine whether the spatial coordinates of the candidate micro-weather station are within the projection trajectory. If so, determine the corresponding candidate micro-weather station as a shadow micro-weather station.
[0148] Among them, candidate micro-weather stations are all weather stations to be determined within the array area, and shaded micro-weather stations are stations whose coordinates fall inside or on the boundary of the banded envelope.
[0149] Understandably, after generating the geometric model of the dynamic sweeping band, the system executes a spatial collision detection algorithm to locate the shadowed micro-weather stations. The system iterates through the absolute geographic coordinates of all candidate micro-weather stations deployed within the photovoltaic array area, substituting them one by one into the boundary equation of the aforementioned strip-shaped envelope surface to determine their positional relationships. If the coordinates of a candidate micro-weather station are located inside or on the boundary of this strip-shaped envelope surface, the system determines that the station is within the array's shadow coverage area and marks it as a shadowed micro-weather station. In actual operation, the system also calculates the theoretical timestamp of the expected arrival of the light spot's shadow at each shadowed micro-weather station based on the speed of the light spot's movement and the linear spatial distance between each shadowed micro-weather station and the starting point.
[0150] For example, in the case of a mountainous photovoltaic power station with significant terrain undulations, the two-dimensional planar swept-strip model may produce spatial projection deviations. In this case, the system introduces digital elevation model data to upgrade the two-dimensional strip-shaped envelope surface into a three-dimensional curved surface swept channel that conforms to the terrain undulations. The system extracts the elevation information of each candidate micro-weather station and calculates its relative positional relationship with the three-dimensional curved surface swept channel in three-dimensional space. Through this three-dimensional spatial topology matching, the system identifies those micro-weather stations whose projection is covered on the two-dimensional plane, but which are actually not truly blocked by the light spots due to their location in low-lying terrain, and removes these stations from the set of shaded micro-weather stations.
[0151] S603: Based on the movement speed of the isolated light spot and the spatial coordinates of the shadow micro-weather station, calculate the coverage time period of the isolated light spot covering the shadow micro-weather station, and extract the real-time irradiance sequence of the shadow micro-weather station according to the coverage time period to obtain the shadow irradiance sequence.
[0152] The coverage time period is the time interval from when the light spot approaches and completely blocks the shadow site until it leaves the shadow site. The shadow irradiance sequence is a set of irradiance data records that have been dynamically clipped by a time window and aligned with the zero point.
[0153] It should be noted that after identifying all the shadow micro-weather stations, the system retrieves historical and real-time monitoring data from these stations via the underlying data bus to obtain the shadow irradiance sequence of each station. Based on the previously calculated expected arrival theoretical timestamp, the system sets a dynamic time-intercepting window that includes both pre- and post-pre-constituent margins. Furthermore, the system utilizes this dynamic time-intercepting window to precisely trim the raw irradiance data stream continuously uploaded by the shadow micro-weather stations. The extracted shadow irradiance sequence contains data records of the complete physical process of the shadow patch moving from approaching, completely obscuring, to gradually moving away from the station. The system performs a relative zero-point alignment operation on all acquired shadow irradiance sequences, that is, uniformly shifting the data points in each sequence where irradiance begins to decrease significantly to the starting position of the relative time coordinate system, forming a set of spatially distributed but highly synchronized shadow irradiance sequences in terms of temporal evolution logic.
[0154] This embodiment accurately predicts and locks micro-weather stations within the array that will be swept by light spot shadows by constructing a dynamic sweeping geometric model and three-dimensional spatial collision detection. At the same time, by using dynamic time window clipping and time axis alignment operations, a highly synchronized and complete set of shadow irradiance sequences is obtained, effectively overcoming the projection deviation caused by terrain undulations.
[0155] In an exemplary embodiment, the irradiance of the target array region is reconstructed based on the target station compression weights and the shadow irradiance sequence to obtain a fused irradiance sequence, including:
[0156] S701: Update the initial station weights of the target micro-weather station according to the target station compression weights. Based on the updated micro-weather station weights and the shadow irradiance sequence, perform spatial interpolation on the target array region to obtain the reconstructed input sequence.
[0157] Among them, the initial station weight is the basic weight determined by the sensor accuracy and historical reliability, the updated micro-weather station weight is the dynamically adjusted weight after fusion and compression, the spatial interpolation is an algorithm that transforms discrete station observations into a continuous planar distribution matrix, and the reconstructed input sequence is a stack of distribution matrices containing complete spatiotemporal evolution information.
[0158] Specifically, after acquiring the shading irradiance sequences of each shading micro-weather station, the system immediately enters the spatial data fusion and reconstruction stage. The system first updates the weights of the target micro-weather stations according to their compression weights. These target station compression weights are essentially dynamic penalty coefficients calculated based on spatial information redundancy. In distributed photovoltaic arrays, if the spatial layout of some micro-weather stations is too dense, the irradiance data they collect during spot scanning often exhibit high similarity. To avoid this data redundancy causing local overfitting in subsequent calculations, the system assigns these stations a smaller compression weight.
[0159] In one possible implementation, the system extracts the initial basic weights of the target micro-weather stations, which are typically determined by the station's sensor accuracy and historical operational reliability. Subsequently, the system fuses these initial basic weights with the calculated target station compression weights, for example, through a multiplication factor adjustment, to calculate the updated weights of each micro-weather station. After this update process, micro-weather stations located in the core shadow region and with strong data independence are assigned higher weights, while the weights of stations on the edge or with high data overlap are correspondingly reduced, thus ensuring the balance and representativeness of the entire array's sensing data. Based on the above steps, the system further performs spatial interpolation reconstruction on the target array region according to the updated weights of each micro-weather station and the shadow irradiance sequence, obtaining a reconstructed input sequence. The purpose of spatial interpolation reconstruction is to transform the point-like monitoring data of discretely distributed micro-weather stations into continuous planar data that can cover the entire physical range of the photovoltaic array, so as to truly restore the spatial distribution of light spot shadows on the array.
[0160] Specifically, for the spatial interpolation reconstruction process, the system first discretizes the vast target array area into several virtual micro-photovoltaic grid units of equal size. For each time step in the shading irradiance sequence, the system extracts the instantaneous irradiance observations from all shading micro-weather stations at that specific moment. Subsequently, the system employs a spatial interpolation strategy incorporating weighting factors to calculate the estimated irradiance for each grid unit. The system calculates the spatial straight-line distance from the target grid unit to all shading micro-weather stations and constructs a distance decay function. In this function, the system not only uses spatial distance as an inverse decay term, but more importantly, it substitutes the previously updated weights of each micro-weather station as a positive adjustment factor. This means that the closer a micro-weather station is to the target grid unit and the higher its updated weight, the greater its contribution to the estimated irradiance of that grid unit. The system traverses all virtual micro-photovoltaic grid units within the target array area, performing the above interpolation calculations one by one, thereby generating the array spatial irradiance distribution matrix for that time step. As the timeline advances, the system stacks the distribution matrices of multiple consecutive time steps in chronological order, ultimately constructing a reconstructed input sequence containing complete spatiotemporal evolution information.
[0161] For example, in the scenario of a mountainous photovoltaic power station with complex terrain, conventional two-dimensional planar distance calculations may lead to interpolation distortion. In this case, the system introduces a three-dimensional spatial elevation difference as an additional distance penalty term during the spatial interpolation reconstruction process. If there is a significant elevation difference between a virtual micro-photovoltaic grid cell and a neighboring micro-weather station, such as one located on a ridge and the other in a deep ravine, the system will significantly amplify the equivalent spatial distance between them in the distance attenuation function. This approach effectively avoids erroneously interpolating the strong sunlight data from high-altitude stations to low-lying grids in the terrain's shadow, ensuring the physical authenticity of the reconstructed input sequence in complex terrain.
[0162] S702: Calculate the coefficient of variation and skewness coefficient based on the reconstructed input sequence. If the coefficient of variation is greater than the preset threshold, construct statistical features based on the coefficient of variation and skewness coefficient.
[0163] Among them, the coefficient of variation is a statistical measure that quantifies the degree of uneven illumination within the array, the skewness coefficient is a statistical measure that reflects the direction and degree of skewness in the irradiance distribution, and the statistical features are a data set that characterizes the dynamic influence of light spots from multiple dimensions such as overall intensity, local fluctuations, and extreme occlusion ratios.
[0164] It should be noted that after successfully constructing the high-dimensional reconstructed input sequence, the system needs to perform dimensionality reduction and feature extraction. Based on the reconstructed input sequence, the system extracts irradiance statistical feature values from it to obtain a fused irradiance sequence. The process of extracting irradiance statistical feature values involves deep analysis of the spatial distribution matrix. For each time step matrix in the reconstructed input sequence, the system performs multi-dimensional statistical operations. First, the system calculates the arithmetic mean of the irradiance of all grid cells as the global average irradiance feature, reflecting the overall light reception level of the photovoltaic array at that moment. Simultaneously, the system calculates the spatial standard deviation of the irradiance of the grid cells; this feature value can accurately quantify the degree of uneven illumination within the array caused by light spot shading. Furthermore, the system sets an irradiance critical threshold corresponding to the photovoltaic inverter's startup power, and the proportion of heavily shaded grid cells in the statistical matrix below this threshold. These carefully designed irradiance statistical feature values comprehensively characterize the dynamic impact of isolated light spots on the photovoltaic array from multiple dimensions, including overall intensity, local fluctuations, and extreme shading ratios.
[0165] S703: The statistical features are fused with the reconstructed input sequence to obtain the fused irradiance sequence.
[0166] The fused irradiance sequence is a compact feature vector containing the absolute shading intensity and the spatial illumination gradient variation law, which is used as the main input of the power prediction model.
[0167] Preferably, the system rigorously aligns and concatenates the multidimensional irradiance statistical feature values extracted at each time step according to the chronological order of the time series. Through this deep fusion of spatiotemporal features, the originally large and loosely packed grid matrix data is transformed into a set of compact and information-dense fused irradiance sequences. This set of sequences not only includes the absolute occlusion intensity of the light spot shadows but also implicitly contains the spatial illumination gradient variation law caused by shadow movement. Finally, the system uses this high-quality fused irradiance sequence as the core feature vector and inputs it into the backend distributed photovoltaic power prediction model, thus completely completing the input construction for distributed photovoltaic power prediction. This process provides the prediction model with pre-process data support that closely matches the actual shading physics, enabling the model to accurately capture the nonlinear drop and recovery trend of photovoltaic array output power caused by local light spot occlusion.
[0168] This embodiment performs spatial interpolation reconstruction by fusing target station compression weights and introduces a three-dimensional elevation penalty to avoid terrain distortion. Subsequently, multi-dimensional statistical feature extraction and spatiotemporal fusion are performed on the reconstruction matrix, successfully transforming the massive and loose grid data into a compact and dense fused irradiance sequence. This provides the power prediction model with input support that closely matches the actual physical process, enabling the model to accurately capture the nonlinear power fluctuation trend caused by local spot occlusion.
[0169] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.
[0170] Based on the same inventive concept, this application also provides a distributed photovoltaic power prediction device based on a micro-weather station for implementing the distributed photovoltaic power prediction method based on a micro-weather station as described above. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations in the embodiments of the distributed photovoltaic power prediction device based on a micro-weather station provided below can be found in the limitations of the distributed photovoltaic power prediction method based on a micro-weather station described above, and will not be repeated here.
[0171] Please see Figure 2 This invention also provides a distributed photovoltaic power prediction device based on a micro-weather station, comprising:
[0172] The irradiance gradient analysis module is used to perform spatial gradient analysis on the spatial distribution of irradiance between adjacent micro-weather stations and obtain a sequence of spatial gradient analysis results.
[0173] The target micro-weather station acquisition module is used to determine the occurrence of an isolated light spot crossing event if the spatial gradient analysis result sequence indicates that there is a spatial abrupt change in irradiance, and to obtain the target micro-weather station where the isolated light spot crossing event occurred.
[0174] The weight compression module is used to obtain the initial near-station weights of the target micro-weather station and compress the initial near-station weights to obtain the compressed weights of the target station.
[0175] The shadow irradiance sequence acquisition module is used to determine the shadow micro-weather stations within the array shadow coverage area based on the movement direction and speed of isolated light spots, and to obtain the shadow irradiance sequence of the shadow micro-weather stations.
[0176] The fused irradiance sequence generation module is used to reconstruct the irradiance of the target array region based on the target station compression weight and the shadow irradiance sequence to obtain the fused irradiance sequence.
[0177] The photovoltaic power prediction module is used to predict photovoltaic power based on the fused irradiance sequence and obtain the prediction results.
[0178] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.
[0179] Reference Figure 3 This invention also provides a computer device, including: a memory and a processor, and a computer program stored in the memory. When the computer program is executed on the processor, it performs the following steps:
[0180] Spatial gradient analysis was performed on the spatial distribution of irradiance between adjacent micro-weather stations to obtain a sequence of spatial gradient analysis results.
[0181] If the spatial gradient analysis result sequence indicates a spatial abrupt change in irradiance, then an isolated spot crossing event is determined to have occurred, and the target micro-weather station where the isolated spot crossing event occurred is obtained.
[0182] Obtain the initial near-station weights of the target micro-weather station, and compress the initial near-station weights to obtain the compressed weights of the target station;
[0183] Based on the direction and speed of movement of isolated light spots, the shadow micro-weather stations within the array shadow coverage area are identified, and the shadow irradiance sequence of the shadow micro-weather stations is obtained.
[0184] Based on the target station compression weight and the shadow irradiance sequence, the irradiance of the target array region is reconstructed to obtain the fused irradiance sequence;
[0185] Photovoltaic power prediction was performed based on the fused irradiance sequence, and the prediction results were obtained.
[0186] The computer device may be a desktop computer, laptop, handheld computer, or cloud server, etc. This computer device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that... Figure 3 The examples of computer devices are merely examples and do not constitute a limitation on computer devices. They may include more or fewer components than shown in the illustration, or combinations of certain components, or different components. For example, they may also include input / output devices, network access devices, etc.
[0187] The processor referred to can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.
[0188] In some embodiments, the memory may be an internal storage unit of the computer device, such as a hard drive or RAM. In other embodiments, the memory may be an external storage device of the computer device, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card. Furthermore, the memory may include both internal and external storage units of the computer device. The memory is used to store the operating system, applications, boot loader, data, and other programs, such as the program code of the computer program. The memory can also be used to temporarily store data that has been output or will be output.
[0189] This invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, performs the following steps:
[0190] Spatial gradient analysis was performed on the spatial distribution of irradiance between adjacent micro-weather stations to obtain a sequence of spatial gradient analysis results.
[0191] If the spatial gradient analysis result sequence indicates a spatial abrupt change in irradiance, then an isolated spot crossing event is determined to have occurred, and the target micro-weather station where the isolated spot crossing event occurred is obtained.
[0192] Obtain the initial near-station weights of the target micro-weather station, and compress the initial near-station weights to obtain the compressed weights of the target station;
[0193] Based on the direction and speed of movement of isolated light spots, the shadow micro-weather stations within the array shadow coverage area are identified, and the shadow irradiance sequence of the shadow micro-weather stations is obtained.
[0194] Based on the target station compression weight and the shadow irradiance sequence, the irradiance of the target array region is reconstructed to obtain the fused irradiance sequence;
[0195] Photovoltaic power prediction was performed based on the fused irradiance sequence, and the prediction results were obtained.
[0196] In this embodiment, if the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include at least: any entity or device capable of carrying computer program code to a photographing device / terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium. Examples include USB flash drives, portable hard drives, magnetic disks, or optical disks. In some jurisdictions, according to legislation and patent practice, computer-readable media cannot be electrical carrier signals or telecommunication signals.
[0197] This invention provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:
[0198] Spatial gradient analysis was performed on the spatial distribution of irradiance between adjacent micro-weather stations to obtain a sequence of spatial gradient analysis results.
[0199] If the spatial gradient analysis result sequence indicates a spatial abrupt change in irradiance, then an isolated spot crossing event is determined to have occurred, and the target micro-weather station where the isolated spot crossing event occurred is obtained.
[0200] Obtain the initial near-station weights of the target micro-weather station, and compress the initial near-station weights to obtain the compressed weights of the target station;
[0201] Based on the direction and speed of movement of isolated light spots, the shadow micro-weather stations within the array shadow coverage area are identified, and the shadow irradiance sequence of the shadow micro-weather stations is obtained.
[0202] Based on the target station compression weight and the shadow irradiance sequence, the irradiance of the target array region is reconstructed to obtain the fused irradiance sequence;
[0203] Photovoltaic power prediction was performed based on the fused irradiance sequence, and the prediction results were obtained.
[0204] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.
[0205] Those skilled in the art will recognize that the units 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.
[0206] In the embodiments disclosed in this application, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling or direct coupling or communication connection may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.
[0207] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for predicting distributed photovoltaic power based on micro-weather stations, characterized in that, Includes the following steps: Spatial gradient analysis was performed on the spatial distribution of irradiance between adjacent micro-weather stations to obtain a sequence of spatial gradient analysis results. If the spatial gradient analysis result sequence indicates a spatial abrupt change in irradiance, then an isolated spot crossing event is determined to have occurred, and the target micro-weather station where the isolated spot crossing event occurred is obtained. The initial near-station weights of the target micro-weather station are obtained, and the initial near-station weights are compressed to obtain the compressed weights of the target station. Based on the direction and speed of movement of isolated light spots, the shadow micro-weather stations within the array shadow coverage area are determined, and the shadow irradiance sequence of the shadow micro-weather stations is obtained. Based on the target station compression weight and the shadow irradiance sequence, the irradiance of the target array region is reconstructed to obtain a fused irradiance sequence. Photovoltaic power prediction was performed based on the fused irradiance sequence, and the prediction results were obtained.
2. The distributed photovoltaic power prediction method based on micro-weather stations according to claim 1, characterized in that, Spatial gradient analysis was performed on the spatial distribution of irradiance between adjacent micro-weather stations, yielding a sequence of spatial gradient analysis results, including: The real-time irradiance sequences of multiple micro-weather stations and the spatial coordinates of the multiple micro-weather stations were obtained; The adjacent micro-weather stations are determined based on the spatial coordinates; The irradiance difference is calculated based on the real-time irradiance sequence of the adjacent micro-weather stations. The spatial distance and directional angle are calculated based on the spatial coordinates of the adjacent micro-weather stations. The gradient vector is obtained based on the irradiance difference, the spatial distance, and the directional angle. The gradient vector is used to extract spatial distribution features, and these features are combined according to the time dimension to obtain the spatial gradient analysis result sequence.
3. The distributed photovoltaic power prediction method based on micro-weather stations according to claim 1, characterized in that, If the spatial gradient analysis result sequence indicates a spatial abrupt change in irradiance, an isolated light spot crossing event is determined to have occurred, and the target micro-weather station where the isolated light spot crossing event occurred is obtained, including: Identify spatial abrupt changes in irradiance within the spatial gradient analysis result sequence, and designate the micro-weather stations corresponding to these spatial abrupt changes as abrupt change nodes. The occlusion contour of the isolated light spot is constructed based on the coordinates of the mutation node. If the movement trajectory of the occlusion contour in the continuous time series passes through the preset micro-weather station monitoring area, it is determined that an isolated light spot crossing event has occurred in the preset micro-weather station monitoring area. If the duration of the isolated spot crossing event is greater than a preset time threshold, a spot crossing status identifier is generated, and the target micro-weather station is determined from the mutation nodes based on the spot crossing status identifier.
4. The distributed photovoltaic power prediction method based on micro-weather stations according to claim 1, characterized in that, Obtain the initial near-station weights of the target micro-weather station, and compress the initial near-station weights to obtain the compressed weights of the target station, including: The spatial distance between the target micro-weather station and its neighboring micro-weather stations is taken as the near-station distance. The reciprocal of the near-station distance is calculated to obtain the distance attenuation base. The initial near-station weight is calculated based on the distance attenuation base. If the initial near-station weight is greater than the preset weight threshold, the historical occlusion frequency of the target micro-weather station is obtained as an attenuation factor, and the attenuation factor is multiplied by the initial near-station weight to obtain the weight attenuation amount. A nonlinear mapping function is constructed based on the weight decay amount. The initial near-station weights are then mapped using the nonlinear mapping function to obtain a dynamic compression coefficient. The initial near-station weights are multiplied by the dynamic compression coefficient to obtain the target station compression weight.
5. The distributed photovoltaic power prediction method based on micro-weather stations according to claim 1, characterized in that, The direction and velocity of movement of an isolated light spot are determined based on the time difference of irradiance changes. The determination process includes: Extract the trough times of the real-time irradiance sequences of the target micro-weather station and adjacent micro-weather stations, and determine the time delay based on the temporal difference between the trough times of the target micro-weather station and the trough times of adjacent micro-weather stations. The distance vector is calculated based on the spatial coordinates of the target micro-weather station and adjacent micro-weather stations. The velocity scalar of the isolated light spot is calculated based on the distance vector and the time delay. The orientation angle of the isolated light spot is calculated based on the distance vector. The direction and speed of motion of the isolated light spot are determined based on the velocity scalar and the direction angle.
6. The distributed photovoltaic power prediction method based on micro-weather stations according to claim 1, characterized in that, Based on the movement direction and speed of isolated light spots, the shadow micro-weather stations within the array's shadow coverage area are determined, and the shadow irradiance sequence of the shadow micro-weather stations is obtained, including: Based on the direction and speed of motion of the isolated light spot, and combined with the initial outline of the isolated light spot, the projection trajectory of the isolated light spot in the target array region is determined. Determine whether the spatial coordinates of the candidate micro-weather station are within the projection trajectory. If so, determine the corresponding candidate micro-weather station as the shadow micro-weather station. Based on the movement speed of the isolated light spot and the spatial coordinates of the shadow micro-weather station, the coverage time period of the isolated light spot covering the shadow micro-weather station is calculated. Based on the coverage time period, the real-time irradiance sequence of the shadow micro-weather station is extracted to obtain the shadow irradiance sequence.
7. The distributed photovoltaic power prediction method based on micro-weather stations according to claim 1, characterized in that, Based on the target station compression weights and the shadow irradiance sequence, the irradiance of the target array region is reconstructed to obtain a fused irradiance sequence, including: The initial station weights of the target micro-weather station are updated according to the target station compression weights. Based on the updated micro-weather station weights and the shadow irradiance sequence, spatial interpolation is performed on the target array region to obtain the reconstructed input sequence. The coefficient of variation and skewness coefficient are calculated based on the reconstructed input sequence. If the coefficient of variation is greater than a preset threshold, statistical features are constructed based on the coefficient of variation and the skewness coefficient. The statistical features are fused with the reconstructed input sequence to obtain the fused irradiance sequence.
8. A distributed photovoltaic power prediction device based on a micro-weather station, characterized in that, include: The irradiance gradient analysis module is used to perform spatial gradient analysis on the spatial distribution of irradiance between adjacent micro-weather stations and obtain a sequence of spatial gradient analysis results. The target micro-weather station acquisition module is used to determine the occurrence of an isolated light spot crossing event if the spatial gradient analysis result sequence indicates that there is a spatial abrupt change in irradiance, and to obtain the target micro-weather station where the isolated light spot crossing event occurred. The weight compression module is used to obtain the initial near-station weights of the target micro-weather station and compress the initial near-station weights to obtain the compressed weights of the target station. The shadow irradiance sequence acquisition module is used to determine the shadow micro-weather stations within the array shadow coverage area based on the movement direction and speed of the isolated light spots, and to obtain the shadow irradiance sequence of the shadow micro-weather stations. The fused irradiance sequence generation module is used to reconstruct the irradiance of the target array region according to the target station compression weight and the shadow irradiance sequence to obtain the fused irradiance sequence. A photovoltaic power prediction module is used to predict photovoltaic power based on the fused irradiance sequence and obtain the prediction result.
9. A computer device, characterized in that, The device includes a processor and a memory: The memory is used to store computer programs and send the instructions of the computer programs to the processor; The processor executes the distributed photovoltaic power prediction method based on micro-weather stations as described in any one of claims 1-7 according to the instructions of the computer program.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the distributed photovoltaic power prediction method based on micro-weather stations as described in any one of claims 1-7.
