A distributed photovoltaic ultra-short-term power prediction method and device based on irradiance saddle point dynamic correction

Abstract

This invention relates to the field of new energy power prediction technology, specifically providing a method and apparatus for distributed photovoltaic (PV) ultra-short-term power prediction based on dynamic correction of irradiance saddle points. The method includes: determining the migration vector of irradiance saddle points in a region based on a time series of regional irradiance contour maps; constructing a cross-correlation function between power plants based on their irradiance time series, and solving the cross-correlation function with the objective of maximizing its value to obtain the cross-correlation delay between power plants in the region; determining a phase correction factor based on the migration vector of the irradiance saddle points and the cross-correlation delay between power plants in the region, and determining the distributed PV power prediction result based on the phase correction factor. The technical solution provided by this invention can significantly alleviate regional phase misalignment, significantly reduce errors, maintain stable correction effects in multi-saddle point and complex terrain scenarios, and support local lightweight operation and remote parameter hot updates.

Description

Technical Field

[0001] This invention relates to the field of new energy power prediction technology, specifically to a method and device for predicting distributed photovoltaic ultra-short-term power based on dynamic correction of irradiance saddle point. Background Technology With the large-scale grid connection of distributed photovoltaic power, minute-level ultra-short-term power forecasting is of key significance for grid dispatch optimization, new energy consumption security, and power plant operation and maintenance efficiency improvement.

[0002] However, due to the combined effects of rapid cloud movement, terrain shading, and ground albedo, the regional irradiance exhibits strong spatial non-stationary propagation characteristics and phase misalignment: the peak and valley irradiance values ​​of adjacent photovoltaic power plants often have time deviations of several minutes, causing traditional prediction models based on single-point time-series data to frequently experience peak misalignment and amplified amplitude deviations. Existing prediction methods often lack modeling of the direction and speed of irradiance propagation, making it difficult to capture the dynamic changes in regional phase errors in real time and to make timely and targeted corrections. This severely limits the accuracy and robustness of ultra-short-term power prediction and makes it difficult to meet the actual needs of refined power grid dispatching. Summary of the Invention

[0003] To overcome the above-mentioned shortcomings, this invention proposes a method and device for predicting distributed photovoltaic ultra-short-term power based on dynamic correction of irradiance saddle point.

[0004] Firstly, a method for predicting the ultra-short-term power of distributed photovoltaic (PV) systems based on dynamic correction of irradiance saddle points is provided. This method includes: The migration vector of the irradiance saddle point in the region is determined based on the time series of the regional irradiance contour map. Based on the irradiance time series of each station in the region, a cross-correlation function between each station is constructed, and the cross-correlation function is solved with the objective of maximizing the cross-correlation function value to obtain the cross-correlation delay between each station in the region. The phase correction factor is determined based on the migration vector of the irradiance saddle point in the region and the cross-correlation delay between the stations in the region, and the distributed photovoltaic power prediction result is determined based on the phase correction factor.

[0005] Preferably, the determination of the migration vector of the irradiance saddle point in a region based on the time series of the regional irradiance contour map includes: Gradient field calculations are performed on the irradiance contour map to obtain the gradient magnitude and gradient direction of each pixel in the irradiance contour map; A set of potential saddle points is constructed based on the gradient magnitude and gradient direction of each pixel; The nonmaximum suppression algorithm is used to remove redundant points from the potential saddle point set; Saddle points are selected from the potential saddle point set based on the confidence of pixels in the potential saddle point set, and a saddle point set is constructed. Cross-frame saddle point pairing is performed on the saddle points of the irradiance contour map of each frame in the time series, and the migration vector of each saddle point is obtained.

[0006] Furthermore, the migration vector includes at least one of the following: propagation speed, propagation direction, and confidence level.

[0007] Furthermore, the nearest neighbor matching algorithm or the Hungarian matching algorithm is used to perform cross-frame saddle point pairing for the saddle points of the irradiance contour map of each frame in the time series.

[0008] Furthermore, the construction of a set of potential saddle points based on the gradient magnitude and gradient direction of each pixel includes: If the gradient magnitude of a pixel is the cross extreme value of the gradient magnitudes within its 3×3 neighborhood, and the gradient direction difference between the pixel and at least two of its 8 neighborhood pixels has a direction difference greater than a preset threshold, then the pixel is added to the potential saddle point set; otherwise, the operation ends.

[0009] Furthermore, the step of filtering saddle points in the potential saddle point set based on the confidence level of pixels in the potential saddle point set includes: If the confidence level of a pixel exceeds the confidence threshold, then the pixel is designated as a saddle point; otherwise, the operation ends.

[0010] Furthermore, the confidence level of the pixel is as follows:

[0011] In the above formula, For the confidence level of a pixel, The maximum gradient magnitude within a 3×3 pixel neighborhood. It is the minimum gradient magnitude within a 3×3 pixel neighborhood. It represents the average gradient magnitude within a 3×3 pixel neighborhood.

[0012] Preferably, the cross-correlation function between the various stations is as follows:

[0013] In the above formula, Here is the irradiance data for station i at time t. Here is the irradiance data for station j at time t. for Irradiance data at station j at time point. For the calculation period, For cross-correlation delay, Let be the cross-correlation function between station i and station j.

[0014] Furthermore, the phase correction factor is as follows:

[0015] In the above formula, For phase correction factor, To integrate weights, This represents the total number of stations within the region. Weighted by the spatial distance between station i and station j. The cross-correlation delay between station i and station j The propagation speed of the saddle point in the migration vector. Let be the vertical distance from station i to the irradiation propagation path. The direction of propagation of the saddle point in the migration vector. The azimuth angle of station i relative to the cluster center.

[0016] Furthermore, the distributed photovoltaic power prediction result is as follows: in, The result is the standard prediction for station i at time t. For dynamic weights.

[0017] Secondly, a distributed photovoltaic ultra-short-term power prediction device based on dynamic correction of irradiance saddle point is provided, the distributed photovoltaic ultra-short-term power prediction device based on dynamic correction of irradiance saddle point includes: The first analysis module is used to determine the migration vector of the irradiance saddle point in the region based on the time series of the regional irradiance contour map; The second analysis module is used to construct the cross-correlation function between each station based on the irradiance time series of each station in the region, and to solve the cross-correlation function with the objective of maximizing the cross-correlation function value, so as to obtain the cross-correlation delay between each station in the region. The third analysis module is used to determine the phase correction factor based on the migration vector of the irradiance saddle point in the region and the cross-correlation delay between the stations in the region, and to determine the distributed photovoltaic power prediction result based on the phase correction factor.

[0018] Preferably, the determination of the migration vector of the irradiance saddle point in a region based on the time series of the regional irradiance contour map includes: Gradient field calculations are performed on the irradiance contour map to obtain the gradient magnitude and gradient direction of each pixel in the irradiance contour map; A set of potential saddle points is constructed based on the gradient magnitude and gradient direction of each pixel; The nonmaximum suppression algorithm is used to remove redundant points from the potential saddle point set; Saddle points are selected from the potential saddle point set based on the confidence of pixels in the potential saddle point set, and a saddle point set is constructed. Cross-frame saddle point pairing is performed on the saddle points of the irradiance contour map of each frame in the time series, and the migration vector of each saddle point is obtained.

[0019] Furthermore, the migration vector includes at least one of the following: propagation speed, propagation direction, and confidence level.

[0020] Furthermore, the nearest neighbor matching algorithm or the Hungarian matching algorithm is used to perform cross-frame saddle point pairing for the saddle points of the irradiance contour map of each frame in the time series.

[0021] Furthermore, the construction of a set of potential saddle points based on the gradient magnitude and gradient direction of each pixel includes: If the gradient magnitude of a pixel is the cross extreme value of the gradient magnitudes within its 3×3 neighborhood, and the gradient direction difference between the pixel and at least two of its 8 neighborhood pixels has a direction difference greater than a preset threshold, then the pixel is added to the potential saddle point set; otherwise, the operation ends.

[0022] Furthermore, the step of filtering saddle points in the potential saddle point set based on the confidence level of pixels in the potential saddle point set includes: If the confidence level of a pixel exceeds the confidence threshold, then the pixel is designated as a saddle point; otherwise, the operation ends.

[0023] Furthermore, the confidence level of the pixel is as follows:

[0024] In the above formula, For the confidence level of a pixel, The maximum gradient magnitude within a 3×3 pixel neighborhood. It is the minimum gradient magnitude within a 3×3 pixel neighborhood. It represents the average gradient magnitude within a 3×3 pixel neighborhood.

[0025] Preferably, the cross-correlation function between the various stations is as follows:

[0026] In the above formula, Here is the irradiance data for station i at time t. Here is the irradiance data for station j at time t. for Irradiance data at station j at time point. For the calculation period, For cross-correlation delay, Let be the cross-correlation function between station i and station j.

[0027] Furthermore, the phase correction factor is as follows:

[0028] In the above formula, For phase correction factor, To integrate weights, This represents the total number of stations within the region. Weighted by the spatial distance between station i and station j. The cross-correlation delay between station i and station j The propagation speed of the saddle point in the migration vector. Let be the vertical distance from station i to the irradiation propagation path. The direction of propagation of the saddle point in the migration vector. The azimuth angle of station i relative to the cluster center.

[0029] Furthermore, the distributed photovoltaic power prediction result is as follows: in, The result is the standard prediction for station i at time t. For dynamic weights.

[0030] Thirdly, a computer device is provided, comprising: one or more processors; The processor is used to store one or more programs; When the one or more programs are executed by the one or more processors, the distributed photovoltaic ultra-short-term power prediction method based on dynamic correction of irradiance saddle point is implemented.

[0031] Fourthly, a computer-readable storage medium is provided, on which a computer program is stored, wherein when the computer program is executed, the distributed photovoltaic ultra-short-term power prediction method based on dynamic correction of irradiance saddle point is implemented.

[0032] The above-described technical solutions of the present invention have at least one or more of the following beneficial effects: This invention provides a method and apparatus for predicting distributed photovoltaic (PV) ultra-short-term power based on dynamic correction of irradiance saddle points. The method includes: determining the migration vector of irradiance saddle points in a region based on a time series of regional irradiance contour maps; constructing a cross-correlation function between stations based on the irradiance time series of each station in the region, and solving the cross-correlation function with the objective of maximizing its value to obtain the cross-correlation delay between stations in the region; determining a phase correction factor based on the migration vector of the irradiance saddle points and the cross-correlation delay between stations in the region, and determining the distributed PV power prediction result based on the phase correction factor. The technical solution provided by this invention, through joint modeling of saddle point migration vectors and delay matrices, explicitly represents the direction / velocity of irradiance propagation, significantly alleviating regional phase misalignment. In rapidly changing cloudy scenarios, the peak alignment error is shortened and significantly reduced within a minute-level window. Simultaneously, it maintains stable correction effects in scenarios with multiple saddle points and complex terrain, and supports local lightweight operation and remote parameter hot updates. Attached Figure Description

[0033] Figure 1 This is a schematic diagram of the main steps of the distributed photovoltaic ultra-short-term power prediction method based on dynamic correction of irradiance saddle point according to an embodiment of the present invention. Detailed Implementation

[0034] The specific embodiments of the present invention will be further described in detail below with reference to the accompanying drawings.

[0035] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0036] Example 1 See appendix Figure 1 , Figure 1 This is a schematic flowchart illustrating the main steps of a distributed photovoltaic ultra-short-term power prediction method based on dynamic correction of irradiance saddle point, according to an embodiment of the present invention. Figure 1 As shown, the distributed photovoltaic ultra-short-term power prediction method based on dynamic correction of irradiance saddle point in this embodiment of the invention mainly includes the following steps: Step S101: Determine the migration vector of the irradiance saddle point in the region based on the time series of the regional irradiance contour map; Step S102: Construct the cross-correlation function between each station based on the irradiance time series of each station in the region, and solve the cross-correlation function with the objective of maximizing the cross-correlation function value to obtain the cross-correlation delay between each station in the region; Step S103: Determine the phase correction factor based on the migration vector of the irradiance saddle point in the region and the cross-correlation delay between the stations in the region, and determine the distributed photovoltaic power prediction result based on the phase correction factor.

[0037] In this embodiment, the process of obtaining the time series of the irradiance contour map can be as follows: For irradiance data collected from multiple measurement points within a distributed photovoltaic cluster or irradiance data retrieved from satellites, spatiotemporal alignment, missing value interpolation, and data normalization preprocessing are performed. The study area is divided into a fixed-size grid (e.g., 500m×500m, which can be adjusted according to the actual scene), and Kriging interpolation is used to interpolate the preprocessed data to form a continuous and complete irradiance field. Irradiance contour maps are generated by setting a fixed contour interval, and a time-series contour map sequence is continuously generated at a frame rate of 1 minute to provide standardized input data for subsequent image visual processing.

[0038] In this embodiment, the determination of the migration vector of the irradiance saddle point of a region based on the time series of the regional irradiance contour map includes: Gradient field calculations are performed on the irradiance contour map to obtain the gradient magnitude and gradient direction of each pixel in the irradiance contour map; A set of potential saddle points is constructed based on the gradient magnitude and gradient direction of each pixel; The nonmaximum suppression algorithm is used to remove redundant points from the potential saddle point set; Saddle points are selected from the potential saddle point set based on the confidence of pixels in the potential saddle point set, and a saddle point set is constructed. Cross-frame saddle point pairing is performed on the saddle points of the irradiance contour map of each frame in the time series, and the migration vector of each saddle point is obtained.

[0039] In one embodiment, the migration vector includes at least one of the following: propagation speed, propagation direction, and confidence level.

[0040] In one implementation, a nearest neighbor matching algorithm or a Hungarian matching algorithm is used to perform cross-frame saddle point pairing for the saddle points of each frame of the irradiance contour map in the time series.

[0041] In one implementation, constructing a set of potential saddle points based on the gradient magnitude and gradient direction of each pixel includes: If the gradient magnitude of a pixel is the cross extreme value of the gradient magnitudes within its 3×3 neighborhood, and the gradient direction difference between the pixel and at least two of its 8 neighborhood pixels has a direction difference greater than a preset threshold, then the pixel is added to the potential saddle point set; otherwise, the operation ends.

[0042] In one implementation, the step of filtering saddle points in the potential saddle point set based on the confidence level of pixels in the potential saddle point set includes: If the confidence level of a pixel exceeds the confidence threshold, then the pixel is designated as a saddle point; otherwise, the operation ends.

[0043] In one implementation, the confidence level of the pixel is as follows:

[0044] In the above formula, For the confidence level of a pixel, The maximum gradient magnitude within a 3×3 pixel neighborhood. It is the minimum gradient magnitude within a 3×3 pixel neighborhood. It represents the average gradient magnitude within a 3×3 pixel neighborhood.

[0045] In this embodiment, the cross-correlation function between the various stations is as follows:

[0046] In the above formula, Here is the irradiance data for station i at time t. Here is the irradiance data for station j at time t. for Irradiance data at station j at time point. For the calculation period, For cross-correlation delay, Let be the cross-correlation function between station i and station j.

[0047] In one implementation, the phase correction factor is as follows:

[0048] In the above formula, For phase correction factor, To integrate weights, This represents the total number of stations within the region. Weighted by the spatial distance between station i and station j. The cross-correlation delay between station i and station j The propagation speed of the saddle point in the migration vector. Let be the vertical distance from station i to the irradiation propagation path. The direction of propagation of the saddle point in the migration vector. The azimuth angle of station i relative to the cluster center.

[0049] In one embodiment, the distributed photovoltaic power prediction result is: in, The result is the standard prediction for station i at time t. For dynamic weights.

[0050] In one specific implementation, based on the method of this invention, a fixed sliding window period is used to update the cross-correlation delay matrix and the saddle point identification threshold; an attention module (including spatial attention and channel attention) is introduced to assign higher weights to high gradient regions in the contour map, enhancing the accuracy and correction stability of saddle point migration identification under complex weather conditions; a center-edge collaborative deployment architecture is adopted: the central node is responsible for global parameter management, model updates and distribution, while the edge nodes locally complete the generation of irradiance contour maps, saddle point identification and rolling correction calculations, achieving lightweight deployment while ensuring real-time performance. Specific implementation details are as follows: (1) Construction of regional irradiance contour maps Taking a distributed photovoltaic cluster along a coast as an application scenario, this cluster includes 10 rooftop / ground-mounted photovoltaic power stations. Irradiance measurement data at a 1-minute resolution were collected from each power station. The collected data underwent spatiotemporal alignment processing, and linear interpolation was used to fill in a small number of missing values. The data was then normalized to the [0,1] interval. The cluster coverage area was divided into a fixed grid of 500m × 500m, and Kriging interpolation was used to interpolate the preprocessed data to form a continuous irradiance field. The contour interval was set to... The system generates irradiance contour maps and generates a time-series contour map sequence at a frame rate of 1 minute.

[0051] (2) Saddle point identification and migration feature extraction The Sobel operator is used to calculate the gradient field of each frame of contour map, and the local extremum detection threshold is set to [value missing]. With a directional change threshold of 30 degrees, potential saddle points were initially identified. Adjacent duplicate pseudo-saddle points were eliminated using a non-maximum suppression algorithm, and a candidate set of saddle points with a confidence level higher than 0.8 was retained. In adjacent time frames, the Hungarian matching algorithm was used to complete saddle point pairing and calculate the saddle point migration vector. It was detected that the low-irradiance saddle points in this scene mainly migrate from northeast to southwest, with an average propagation speed of about 1.5 km / min.

[0052] (3) Construction of cross-correlation delay matrix and phase correction factor Cross-correlation functions were calculated for the irradiance sequences of 10 power stations. The time delay corresponding to the maximum correlation between each power station was extracted, and a 10×10 cross-correlation delay matrix was constructed. The delay τ≈8min corresponding to the peak cross-correlation between adjacent stations was calculated. The saddle point migration speed was 1.5 km / min, the migration direction was 225 degrees (northeast to southwest), and the straight-line distance between the stations were integrated. A piecewise linear mapping function was used to generate a phase correction factor.

[0053] (4) Rolling forecast correction mechanism An LSTM model was selected as the baseline predictor to output preliminary prediction results. The initial value of the dynamic weight was set to 0.6, and it was adaptively updated based on the predicted MAE and peak alignment over the past 10 minutes, with an update cycle of 1 minute. When a saddle point was detected to be approaching a power station, the prediction time window for that power station was moved forward by 2-3 minutes, and the dynamic weight was increased to 0.8-1.0. When the saddle point was moved away, the prediction time window was moved backward by 1-2 minutes, and the dynamic weight was adjusted to 0.3-0.5 to achieve minute-level rolling correction.

[0054] (5) Online self-learning and parameter updates The sliding window length is set to 7 days, and the cross-correlation delay matrix and saddle point identification threshold are automatically updated every 7 days. A spatial attention module is introduced to address gradient values ​​in contour maps that are higher than... The region is assigned a weight of 1.2 times; a center-edge collaborative deployment is adopted: the central node is deployed in the cluster management center, which is responsible for parameter updates and model distribution; the edge nodes are deployed locally in each power station to complete contour map generation, saddle point identification and rolling correction calculation, ensuring the real-time performance of data processing.

[0055] After applying the solution in this embodiment, the overall MAE of the ultra-short-term power prediction of the distributed photovoltaic cluster decreased from 12.3% to 8.1%, and the peak phase error was shortened from ±9min to ±3min, with both prediction accuracy and stability being significantly improved.

[0056] Example 2 Based on the same inventive concept, this invention also provides a distributed photovoltaic ultra-short-term power prediction device based on dynamic correction of irradiance saddle point, the distributed photovoltaic ultra-short-term power prediction device based on dynamic correction of irradiance saddle point includes: The first analysis module is used to determine the migration vector of the irradiance saddle point in the region based on the time series of the regional irradiance contour map; The second analysis module is used to construct the cross-correlation function between each station based on the irradiance time series of each station in the region, and to solve the cross-correlation function with the objective of maximizing the cross-correlation function value, so as to obtain the cross-correlation delay between each station in the region. The third analysis module is used to determine the phase correction factor based on the migration vector of the irradiance saddle point in the region and the cross-correlation delay between the stations in the region, and to determine the distributed photovoltaic power prediction result based on the phase correction factor.

[0057] Preferably, the determination of the migration vector of the irradiance saddle point in a region based on the time series of the regional irradiance contour map includes: Gradient field calculations are performed on the irradiance contour map to obtain the gradient magnitude and gradient direction of each pixel in the irradiance contour map; A set of potential saddle points is constructed based on the gradient magnitude and gradient direction of each pixel; The nonmaximum suppression algorithm is used to remove redundant points from the potential saddle point set; Saddle points are selected from the potential saddle point set based on the confidence of pixels in the potential saddle point set, and a saddle point set is constructed. Cross-frame saddle point pairing is performed on the saddle points of the irradiance contour map of each frame in the time series, and the migration vector of each saddle point is obtained.

[0058] Furthermore, the migration vector includes at least one of the following: propagation speed, propagation direction, and confidence level.

[0059] Furthermore, the nearest neighbor matching algorithm or the Hungarian matching algorithm is used to perform cross-frame saddle point pairing for the saddle points of the irradiance contour map of each frame in the time series.

[0060] Furthermore, the construction of a set of potential saddle points based on the gradient magnitude and gradient direction of each pixel includes: If the gradient magnitude of a pixel is the cross extreme value of the gradient magnitudes within its 3×3 neighborhood, and the gradient direction difference between the pixel and at least two of its 8 neighborhood pixels has a direction difference greater than a preset threshold, then the pixel is added to the potential saddle point set; otherwise, the operation ends.

[0061] Furthermore, the step of filtering saddle points in the potential saddle point set based on the confidence level of pixels in the potential saddle point set includes: If the confidence level of a pixel exceeds the confidence threshold, then the pixel is designated as a saddle point; otherwise, the operation ends.

[0062] Furthermore, the confidence level of the pixel is as follows:

[0063] In the above formula, For the confidence level of a pixel, The maximum gradient magnitude within a 3×3 pixel neighborhood. It is the minimum gradient magnitude within a 3×3 pixel neighborhood. It represents the average gradient magnitude within a 3×3 pixel neighborhood.

[0064] Preferably, the cross-correlation function between the various stations is as follows:

[0065] In the above formula, Here is the irradiance data for station i at time t. Here is the irradiance data for station j at time t. for Irradiance data at station j at time point. For the calculation period, For cross-correlation delay, Let be the cross-correlation function between station i and station j.

[0066] Furthermore, the phase correction factor is as follows:

[0067] In the above formula, For phase correction factor, To integrate weights, This represents the total number of stations within the region. Weighted by the spatial distance between station i and station j. The cross-correlation delay between station i and station j The propagation speed of the saddle point in the migration vector. Let be the vertical distance from station i to the irradiation propagation path. The direction of propagation of the saddle point in the migration vector. The azimuth angle of station i relative to the cluster center.

[0068] Furthermore, the distributed photovoltaic power prediction result is as follows: in, The result is the standard prediction for station i at time t. For dynamic weights.

[0069] Example 3 Based on the same inventive concept, this invention also provides a computer device, which includes a processor and a memory. The memory stores a computer program, which includes program instructions. The processor executes the program instructions stored in the computer storage medium. The processor may be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing and control core of the terminal, suitable for implementing one or more instructions, specifically suitable for loading and executing one or more instructions in the computer storage medium to implement corresponding method flows or corresponding functions, thereby realizing the steps of the distributed photovoltaic ultra-short-term power prediction method based on dynamic correction of irradiance saddle point in the above embodiments.

[0070] Example 4 Based on the same inventive concept, this invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device used to store programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the computer device and extended storage media supported by the computer device. The computer-readable storage medium provides storage space that stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more computer programs (including program code). It should be noted that the computer-readable storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the steps of the distributed photovoltaic ultra-short-term power prediction method based on dynamic correction of irradiance saddle point in the above embodiments.

[0071] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0072] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0073] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0074] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0075] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A distributed photovoltaic ultra-short-term power prediction method based on irradiance saddle point dynamic correction, characterized in that, The method comprises: determining a migration vector of a radiation intensity saddle point of a region based on a time sequence of a radiation intensity contour map of the region; constructing a cross-correlation function between each station in the region based on a time sequence of radiation intensity of each station in the region, and solving the cross-correlation function with the maximum cross-correlation function value as the target to obtain a cross-correlation delay between each station in the region; determining a phase correction factor based on the migration vector of the radiation intensity saddle point of the region and the cross-correlation delay between each station in the region, and determining a distributed photovoltaic power prediction result based on the phase correction factor.

2. The method of claim 1, wherein, The method comprises: performing gradient field calculation on the radiation intensity contour map to obtain gradient amplitude and gradient direction of each pixel in the radiation intensity contour map; constructing a potential saddle point set based on the gradient amplitude and gradient direction of each pixel; removing redundant points in the potential saddle point set by using a non-maximum suppression algorithm; screening saddle points in the potential saddle point set based on the confidence of each pixel in the potential saddle point set, and constructing a saddle point set; performing cross-frame saddle point pairing on the saddle points of each frame of the radiation intensity contour map in the time sequence, and obtaining a migration vector of each saddle point.

3. The method of claim 2, wherein, The migration vector comprises at least one of the following: propagation speed, propagation direction, and confidence.

4. The method of claim 2, wherein, The cross-frame saddle point pairing is performed on the saddle points of each frame of the radiation intensity contour map in the time sequence by using a nearest neighbor matching algorithm or a Hungarian matching algorithm.

5. The method of claim 2, wherein, The method comprises: if the gradient amplitude of a pixel is a cross extreme value of the gradient amplitude in a 3*3 neighborhood of the pixel and there are at least two regions whose direction difference is greater than a preset threshold in the gradient direction difference between the pixel and the pixels in its 8 regions, the pixel is added to the potential saddle point set, otherwise, the operation is ended.

6. The method of claim 2, wherein, The method comprises: if the confidence of a pixel exceeds a confidence threshold, the pixel is taken as a saddle point, otherwise, the operation is ended.

7. The method of claim 6, wherein, The confidence of the pixel is as follows: In the above formula, is the confidence of the pixel, is the maximum gradient magnitude in a 3x3 neighborhood of the pixel, is the minimum gradient magnitude in a 3x3 neighborhood of the pixel, is the average gradient magnitude in a 3x3 neighborhood of the pixel.

8. The method of claim 1, wherein, The cross-correlation function between each station is as follows: In the above formulae, is the irradiance data of site i at time t, is the irradiance data of site j at time t, is the irradiance data of site j at time t, is the irradiance data of site j at time t, is the calculation period, is the cross-correlation delay, is the cross-correlation function between site i and site j.

9. The method of claim 8, wherein, The phase correction factor is as follows: In the above formula, is a phase correction factor, is a fusion weight, is the total number of stations in the region, is a spatial distance weighting between station i and station j, is a cross-correlation delay between station i and station j, is a propagation speed of the saddle point in the migration vector, is a vertical distance from station i to the irradiation propagation path, is a propagation direction of the saddle point in the migration vector, is an azimuth angle of station i relative to the cluster center.

10. The method of claim 9, wherein, The distributed photovoltaic power prediction result is as follows, wherein is a conventional prediction result of station i at time t, and is a dynamic weight.

11. A distributed photovoltaic ultra-short-term power prediction device based on irradiance saddle point dynamic correction, characterized in that, The device comprises: a first analysis module configured to determine a migration vector of a radiation intensity saddle point of a region based on a time sequence of a radiation intensity contour map of the region; a second analysis module configured to construct a cross-correlation function between each station in the region based on a time sequence of radiation intensity of each station in the region, and solve the cross-correlation function with the maximum cross-correlation function value as the target to obtain a correlation delay between each station in the region; a third analysis module configured to determine a phase correction factor based on the migration vector of the radiation intensity saddle point of the region and a cross-correlation delay between each station in the region, and determine a distributed photovoltaic power prediction result based on the phase correction factor.

12. The apparatus of claim 11, wherein, The device comprises: performing gradient field calculation on the radiation intensity contour map to obtain gradient amplitude and gradient directions of each pixel in the radiation intensity contour map; constructing a potential saddle point set based on gradient amplitude and gradient direction of each pixel; A set of potential saddle points is constructed based on the gradient magnitude and gradient direction of each pixel; The nonmaximum suppression algorithm is used to remove redundant points from the potential saddle point set; Saddle points are selected from the potential saddle point set based on the confidence of pixels in the potential saddle point set, and a saddle point set is constructed. Cross-frame saddle point pairing is performed on the saddle points of the irradiance contour map of each frame in the time series, and the migration vector of each saddle point is obtained.

13. The apparatus of claim 12, wherein, The migration vector includes at least one of the following: propagation speed, propagation direction, and confidence level.

14. The apparatus of claim 12, wherein, The nearest neighbor matching algorithm or the Hungarian matching algorithm is used to perform cross-frame saddle point pairing for the saddle points of the irradiance contour map of each frame in the time series.

15. The apparatus of claim 12, wherein, The construction of a potential saddle point set based on the gradient magnitude and gradient direction of each pixel includes: If the gradient magnitude of a pixel is the cross extreme value of the gradient magnitudes within its 3×3 neighborhood, and the gradient direction difference between the pixel and at least two of its 8 neighborhood pixels has a direction difference greater than a preset threshold, then the pixel is added to the potential saddle point set; otherwise, the operation ends.

16. The apparatus of claim 12, wherein, The process of filtering saddle points from the potential saddle point set based on the confidence level of pixels in the potential saddle point set includes: If the confidence level of a pixel exceeds the confidence threshold, then the pixel is designated as a saddle point; otherwise, the operation ends.

17. The apparatus of claim 16, wherein, The confidence level of the pixel is as follows: In the above formula, is the confidence of the pixel, is the maximum gradient magnitude in a 3x3 field around the pixel, is the minimum gradient magnitude in a 3x3 field around the pixel, is the average gradient magnitude in a 3x3 field around the pixel.

18. The apparatus of claim 11, wherein, The cross-correlation functions between the various stations are as follows: In the above formulae, is the irradiance data for site i at time t, is the irradiance data for site j at time t, is the irradiance data for site j at time t, is the irradiance data for site j at time t, is the calculation period, is the cross-correlation delay, is the cross-correlation function between site i and site j.

19. The apparatus of claim 18, wherein, The phase correction factor is as follows: In the above formula, is a phase correction factor, is a fusion weight, is the total number of stations in the region, is a spatial distance weighting between station i and station j, is a cross-correlation delay between station i and station j, is the propagation speed of the saddle point in the migration vector, is the vertical distance from station i to the irradiation propagation path, is the propagation direction of the saddle point in the migration vector, is the azimuth angle of station i relative to the cluster center.

20. The apparatus of claim 19, wherein, The distributed photovoltaic power prediction result is given by where is the conventional prediction result of station i at time t, and is the dynamic weight.

21. A computer device, comprising: include: One or more processors; The processor is used to execute one or more programs; When the one or more programs are executed by the one or more processors, the distributed photovoltaic ultra-short-term power prediction method based on dynamic correction of irradiance saddle point as described in any one of claims 1 to 10 is implemented.

22. A computer-readable storage medium, characterized in that, It contains a computer program, which, when executed, implements the distributed photovoltaic ultra-short-term power prediction method based on dynamic correction of irradiance saddle point as described in any one of claims 1 to 10.

Images