Photovoltaic cluster short-term power prediction method and system facing sand and dust weather

Abstract

The invention discloses a photovoltaic cluster short-term power prediction method and system facing sand and dust weather, and belongs to the technical field of photovoltaic prediction. The method comprises the following steps: acquiring weather data and power data of a photovoltaic area in sand and dust weather, preprocessing the weather data and the power data to perform sand and dust scene division on the weather data and the power data, and expanding small sample data in the weather data and the power data to obtain a training set; generating an adaptive combined prediction model based on the training set, inputting a test set into the adaptive combined prediction model, and performing photovoltaic short-term power prediction to obtain photovoltaic short-term prediction power; and based on a statistical upscaling algorithm, according to the photovoltaic short-term prediction power, predicting the photovoltaic cluster short-term prediction power. According to the method, the photovoltaic power in extreme weather with sand and dust in a short period can be effectively predicted through model building and a statistical upscaling algorithm.

Description

Technical Field

[0001] The present invention relates to the field of photovoltaic prediction technology, and more specifically, to a method and system for short-term power prediction of a photovoltaic cluster for sandstorm weather. Background Art

[0002] In 2023, more than half of the world's new installed capacity of renewable energy power generation will be installed, and the cumulative installed capacity will be close to 40%. As of the end of April, my country's cumulative installed capacity of wind power and photovoltaic power generation exceeded 1.1 billion kilowatts, an increase of about 38% year-on-year.

[0003] Deserts and wasteland areas are key development areas for photovoltaics, but these areas suffer from sandstorm disasters all year round, resulting in obvious deviations in photovoltaic power forecasts, posing challenges to the safety and stability of the power system. Therefore, in response to sandstorm disasters, providing accurate and timely photovoltaic cluster power forecasts is a key technical means to deal with the uncertainty of the power system, and it is also an inherent requirement for building a new power system. Summary of the invention

[0004] In view of the above problems, the present invention proposes a short-term power prediction method for photovoltaic clusters in sandstorm weather, including:

[0005] Acquire weather data and power data under sandstorm weather in the photovoltaic area, pre-process the weather data and power data to divide the weather data and power data into sandstorm scenes, and expand small sample data in the weather data and power data to obtain a training set;

[0006] Based on the training set, an adaptive combined prediction model is generated, and the test set is input into the adaptive combined prediction model to perform short-term photovoltaic power prediction to obtain short-term photovoltaic predicted power;

[0007] Based on the statistical upscaling algorithm, the short-term predicted power of the photovoltaic cluster is predicted according to the short-term predicted power of photovoltaics.

[0008] Optional weather data, including numerical weather forecast data and dust data.

[0009] Optionally, preprocessing the weather data and the power data to divide the weather data and the power data into dust scenes includes:

[0010] Determine the start and end time of the sandstorm event according to the weather data, calculate the influence of sandstorm concentration on power generation according to the start and end time of the sandstorm event, and divide the sandstorm weather scenes based on the influence of sandstorm concentration on power generation;

[0011] The sandstorm weather scene division is performed;

[0012] Setting a dust concentration threshold, and determining whether the dust will cause a loss in photovoltaic output based on the dust concentration threshold;

[0013] If, starting from a certain moment, the dust concentration increases and exceeds the dust concentration threshold within a preset time period, it is determined to be a dust-affected scene;

[0014] If the dust concentration begins to be less than the dust concentration threshold at a certain moment and does not rebound within a preset time period, it is determined as a scene without dust influence.

[0015] Optionally, small sample data in weather data and power data are expanded to obtain a training set, including:

[0016] The data in the weather data and power data under dust scenes are taken as small samples, and the small samples are expanded using a time series generative adversarial network algorithm. The isolated forest algorithm is used to remove outliers from the expanded small samples to generate expanded sample data. The expanded sample data and the remaining data in the weather data and power data excluding the small sample data are constructed as a training set.

[0017] An optional, pre-built adaptive combined prediction model includes: a pre-established photovoltaic power prediction benchmark model and a pre-established photovoltaic power prediction correction model under extreme dust weather conditions.

[0018] Optionally, the test set is input into the adaptive combined prediction model to perform short-term photovoltaic power prediction to obtain short-term photovoltaic power prediction, including:

[0019] The test set is divided into data affected by sandstorms and normal weather data according to the sandstorm scenarios, and marked. The normal weather data is predicted using the photovoltaic power prediction benchmark model of the adaptive combined prediction model to obtain a first prediction result. The data affected by sandstorms is predicted using the photovoltaic power prediction correction model of the adaptive combined prediction model to obtain a second prediction result. According to the mark, the first prediction result is combined with the second prediction result using a self-use algorithm to predict the short-term predicted power of the photovoltaic site.

[0020] Optionally, based on the statistical upscaling algorithm, the short-term predicted power of the photovoltaic cluster is predicted according to the short-term predicted power of the photovoltaic site, including:

[0021] By sorting points to identify clustering structures, the OPTICS algorithm and the clustering K-Medoids algorithm are used to divide the photovoltaic cluster area into different sub-areas, and benchmark sites are selected for each sub-area. The output power of the benchmark site and the photovoltaic cluster area is calculated, and the average absolute error and correlation coefficient after normalization are calculated. Based on the average absolute error and correlation coefficient, the weight coefficient of the benchmark site is calculated;

[0022] Different weights are assigned to photovoltaics in different sub-regions, and based on the statistical upscaling algorithm, the short-term predicted power of the photovoltaic cluster area is predicted according to the short-term predicted power corresponding to photovoltaics with different weights.

[0023] On the other hand, the present invention proposes a photovoltaic cluster short-term power prediction system for sandstorm weather, comprising:

[0024] A data processing unit is used to obtain weather data and power data under sandstorm weather in the photovoltaic area, pre-process the weather data and power data to divide the weather data and power data into sandstorm scenes, and expand small sample data in the weather data and power data to obtain a training set;

[0025] A prediction unit, used to generate an adaptive combined prediction model based on the training set, input the test set into the adaptive combined prediction model, perform short-term photovoltaic power prediction, and obtain the short-term photovoltaic predicted power;

[0026] The output unit is used to predict the short-term predicted power of the photovoltaic cluster based on the short-term predicted power of the photovoltaic based on the statistical upscaling algorithm.

[0027] Optional weather data, including numerical weather forecast data and dust data.

[0028] Optionally, preprocessing the weather data and the power data to divide the weather data and the power data into dust scenes includes:

[0029] Determine the start and end time of the sandstorm event according to the weather data, calculate the influence of sandstorm concentration on power generation according to the start and end time of the sandstorm event, and divide the sandstorm weather scenes based on the influence of sandstorm concentration on power generation;

[0030] The sandstorm weather scene division is performed;

[0031] Setting a dust concentration threshold, and determining whether the dust will cause a loss in photovoltaic output based on the dust concentration threshold;

[0032] If, starting from a certain moment, the dust concentration increases and exceeds the dust concentration threshold within a preset time period, it is determined to be a dust-affected scene;

[0033] If the dust concentration begins to be less than the dust concentration threshold at a certain moment and does not rebound within a preset time period, it is determined as a scene without dust influence.

[0034] Optionally, small sample data in weather data and power data are expanded to obtain a training set, including:

[0035] The data in the weather data and power data under dust scenes are taken as small samples, and the small samples are expanded using a time series generative adversarial network algorithm. The isolated forest algorithm is used to remove outliers from the expanded small samples to generate expanded sample data. The expanded sample data and the remaining data in the weather data and power data excluding the small sample data are constructed as a training set.

[0036] An optional, pre-built adaptive combined prediction model includes: a pre-established photovoltaic power prediction benchmark model and a pre-established photovoltaic power prediction correction model under extreme dust weather conditions.

[0037] Optionally, the test set is input into the adaptive combined prediction model to perform short-term photovoltaic power prediction to obtain short-term photovoltaic power prediction, including:

[0038] The test set is divided into data affected by sandstorms and normal weather data according to the sandstorm scenarios, and marked. The normal weather data is predicted using the photovoltaic power prediction benchmark model of the adaptive combined prediction model to obtain a first prediction result. The data affected by sandstorms is predicted using the photovoltaic power prediction correction model of the adaptive combined prediction model to obtain a second prediction result. According to the mark, the first prediction result is combined with the second prediction result using a self-use algorithm to predict the short-term predicted power of the photovoltaic site.

[0039] Optionally, based on the statistical upscaling algorithm, the short-term predicted power of the photovoltaic cluster is predicted according to the short-term predicted power of the photovoltaic site, including:

[0040] By sorting points to identify clustering structures, the OPTICS algorithm and the clustering K-Medoids algorithm are used to divide the photovoltaic cluster area into different sub-areas, and benchmark sites are selected for each sub-area. The output power of the benchmark site and the photovoltaic cluster area is calculated, and the average absolute error and correlation coefficient after normalization are calculated. Based on the average absolute error and correlation coefficient, the weight coefficient of the benchmark site is calculated;

[0041] Different weights are assigned to photovoltaics in different sub-regions, and based on the statistical upscaling algorithm, the short-term predicted power of the photovoltaic cluster area is predicted according to the short-term predicted power corresponding to photovoltaics with different weights.

[0042] In yet another aspect, the present invention further provides a computing device, comprising: one or more processors;

[0043] a processor for executing one or more programs;

[0044] When the one or more programs are executed by the one or more processors, the above-described method is implemented.

[0045] In yet another aspect, the present invention further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed, the method described above is implemented.

[0046] Compared with the prior art, the present invention has the following beneficial effects:

[0047] The present invention provides a short-term power prediction method for photovoltaic clusters in sandstorm weather, including: obtaining weather data and power data under sandstorm weather in the photovoltaic area, preprocessing the weather data and power data to divide the weather data and power data into sandstorm scenes, and expanding small sample data in the weather data and power data to obtain a training set; based on the training set, generating an adaptive combination prediction model, inputting a test set into the adaptive combination prediction model, performing short-term photovoltaic power prediction, and obtaining short-term photovoltaic predicted power; based on a statistical upscaling algorithm, predicting the short-term photovoltaic predicted power. The present invention can effectively predict the photovoltaic power in extreme weather in the short term by building a model and a statistical upscaling algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 is a flow chart of the method of the present invention;

[0049] Figure 2 A flowchart of an embodiment of the method of the present invention;

[0050] Figure 3 A schematic diagram of an adaptive mechanism of an embodiment of the method of the present invention;

[0051] Figure 4 A comparison diagram of the cluster power prediction strategy and the actual total power of an embodiment of the method of the present invention;

[0052] Figure 5 It is a structural diagram of the system of the present invention. DETAILED DESCRIPTION

[0053] Now, exemplary embodiments of the present invention are described with reference to the accompanying drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. These embodiments are provided to disclose the present invention in detail and completely and to fully convey the scope of the present invention to those skilled in the art. The terms used in the exemplary embodiments shown in the accompanying drawings are not intended to limit the present invention. In the accompanying drawings, the same units / elements are marked with the same reference numerals.

[0054] Unless otherwise specified, the terms (including technical terms) used herein have the commonly understood meanings to those skilled in the art. In addition, it is understood that the terms defined in commonly used dictionaries should be understood to have the same meanings as those in the context of the relevant fields, and should not be understood as idealized or overly formal meanings.

[0055] Embodiment 1:

[0056] The present invention proposes a short-term power prediction method for photovoltaic clusters in dust weather. Figure 1 As shown, including:

[0057] Step 1: obtaining weather data and power data under sandstorm weather in the photovoltaic area, preprocessing the weather data and power data to divide the weather data and power data into sandstorm scenes, and expanding small sample data in the weather data and power data to obtain a training set;

[0058] Step 2: Based on the training set, an adaptive combined prediction model is generated, and the test set is input into the adaptive combined prediction model to perform short-term photovoltaic power prediction to obtain short-term photovoltaic predicted power;

[0059] Step 3: Based on the statistical upscaling algorithm, the short-term predicted power of the photovoltaic cluster is predicted according to the short-term predicted power of the photovoltaic system.

[0060] Among them, weather data includes: numerical weather forecast data and sandstorm data.

[0061] The weather data and the power data are preprocessed to divide the weather data and the power data into dust scenes, including:

[0062] Determine the start and end time of the sandstorm event according to the weather data, calculate the influence of sandstorm concentration on power generation according to the start and end time of the sandstorm event, and divide the sandstorm weather scenes based on the influence of sandstorm concentration on power generation;

[0063] The sandstorm weather scene division is performed;

[0064] Setting a dust concentration threshold, and determining whether the dust will cause a loss in photovoltaic output based on the dust concentration threshold;

[0065] If, starting from a certain moment, the dust concentration increases and exceeds the dust concentration threshold within a preset time period, it is determined to be a dust-affected scene;

[0066] If the dust concentration begins to be less than the dust concentration threshold at a certain moment and does not rebound within a preset time period, it is determined as a scene without dust influence.

[0067] Among them, the small sample data in the weather data and power data are expanded to obtain the training set, including:

[0068] The data in the weather data and power data under dust scenes are taken as small samples, and the small samples are expanded using a time series generative adversarial network algorithm. The isolated forest algorithm is used to remove outliers from the expanded small samples to generate expanded sample data. The expanded sample data and the remaining data in the weather data and power data excluding the small sample data are constructed as a training set.

[0069] Among them, the pre-built adaptive combined prediction model includes: a pre-established photovoltaic power prediction benchmark model and a pre-established photovoltaic power prediction correction model under extreme sandstorm weather.

[0070] The test set is input into the adaptive combined prediction model to perform short-term photovoltaic power prediction, and the short-term photovoltaic power prediction is obtained, including:

[0071] The test set is divided into data affected by sandstorms and normal weather data according to the sandstorm scenarios, and marked. The normal weather data is predicted using the photovoltaic power prediction benchmark model of the adaptive combined prediction model to obtain a first prediction result. The data affected by sandstorms is predicted using the photovoltaic power prediction correction model of the adaptive combined prediction model to obtain a second prediction result. According to the mark, the first prediction result is combined with the second prediction result using a self-use algorithm to predict the short-term predicted power of the photovoltaic site.

[0072] Among them, based on the statistical upscaling algorithm, the short-term predicted power of the photovoltaic cluster is predicted according to the short-term predicted power of the photovoltaic site, including:

[0073] By sorting points to identify clustering structures, the OPTICS algorithm and the clustering K-Medoids algorithm are used to divide the photovoltaic cluster area into different sub-areas, and benchmark sites are selected for each sub-area. The output power of the benchmark site and the photovoltaic cluster area is calculated, and the average absolute error and correlation coefficient after normalization are calculated. Based on the average absolute error and correlation coefficient, the weight coefficient of the benchmark site is calculated;

[0074] Different weights are assigned to photovoltaics in different sub-regions, and based on the statistical upscaling algorithm, the short-term predicted power of the photovoltaic cluster area is predicted according to the short-term predicted power corresponding to photovoltaics with different weights.

[0075] The present invention is further described below with reference to specific cases:

[0076] With the expansion of photovoltaic installed capacity, deserts, Gobi and wasteland have become key development areas for photovoltaic power generation. However, these areas are plagued by sandstorms all year round, which seriously threatens the safe and stable operation of new power systems. Providing accurate photovoltaic power forecasts in a targeted manner will be an effective countermeasure. Under the influence of extreme sandstorm weather, the photovoltaic output in key development areas is greatly reduced, and the prediction accuracy is low due to the scarcity of photovoltaic power, meteorological and sandstorm data samples during this period. Based on this, the method of the present invention is proposed, and its implementation process is as follows: Figure 2 As shown, specifically including:

[0077] Dust scene division and expansion of small sample data:

[0078] After preprocessing the numerical weather forecast data, power data, and dust data, the dust weather scene is divided according to the start and end time of the dust event and the impact of dust concentration on power generation. A dust concentration threshold is set to determine whether the dust will cause a significant loss to photovoltaic output. If the dust concentration suddenly increases from a certain moment and exceeds the threshold in the time period t1 to t n If the dust concentration at a certain moment is less than the threshold and does not rebound within time t, it is judged as a scene with dust impact. All the data of scenes with dust impact are combined into small sample data.

[0079] The small amount of data in a small sample will seriously affect the prediction accuracy and generalization ability of the model. Therefore, the time series generative adversarial networks (TimeGAN) algorithm is used to enhance the data of small samples, and the isolation forest (IForest) algorithm is used to remove outliers from the expanded sample data.

[0080] Power prediction and adaptive combined prediction models:

[0081] The numerical weather forecast, i.e., dust data, is used as feature input, and the actual photovoltaic power is used as label output for modeling.

[0082] Among them, the training set samples of daily weather are used to establish a photovoltaic power prediction benchmark model, and then the small dust samples extracted from the training set are expanded by the TimeGAN algorithm. On this basis, the expanded sample data are used to build a photovoltaic power prediction model under extreme dust weather.

[0083] The dust event scene sequence in the test set meteorological feature sequence is marked and divided, and the photovoltaic power prediction model under dust weather is used to predict the power of the marked time series. At the same time, the benchmark model is used to predict the power of the test set, and the normalized mean absolute error of the predicted power of the two models for the marked dust time series is calculated. If the predicted power error of the two models exceeds the error loss threshold, the adaptive mechanism, such as Figure 3 As shown, the predicted power under dust weather will replace the predicted power of the benchmark model in this period, otherwise no switching will be performed.

[0084] Through the above steps, the combined photovoltaic power prediction model can effectively solve the serious problem of photovoltaic power generation loss during daily weather including dust storms and improve the accuracy of power prediction.

[0085] Improved statistical upscaling algorithm to calculate cluster power:

[0086] The cluster area is divided into different sub-areas through OPTICS and K-Medoids algorithms, and benchmark sites are selected for each sub-area. Based on the historical data characteristics, the normalized mean absolute error and correlation coefficient between the benchmark site and the output power of the photovoltaic cluster are calculated, and then the weight coefficient of each benchmark site is calculated. The specific weight coefficient formula can be expressed as:

[0087]

[0088] In the above formula, W i , R i and E i,MAE represents the weight, correlation coefficient and mean absolute error of the i-th benchmark station, w i is the final weight coefficient of the i-th benchmark site.

[0089] According to the weight coefficient calculated in the previous section, different weights are assigned to the photovoltaic sites in each sub-region. The photovoltaic cluster power prediction result formula based on the statistical upscaling algorithm can be expressed as:

[0090] P clu,i =P i *C i2 / C i1 (3)

[0091]

[0092] In the above formula, P i and P clu,i They represent the predicted power of the benchmark site and the predicted power of the sub-regional photovoltaic site, respectively. i1 and C i2 They represent the installed capacity of benchmark sites and the installed capacity of sub-regional PV sites respectively.

[0093] Therefore, based on the above improved statistical upscaling algorithm, the power prediction results of regional PV sites can be calculated from the power prediction results of benchmark sites.

[0094] The present invention combines the predicted power in dust weather with the daily predicted power of photovoltaic power in a targeted manner, which not only improves the accuracy of the predicted power at a single station, but also improves the accuracy of the power prediction of photovoltaic clusters. Figure 4 , as shown in Table 1 and Table 2, where Table 1 shows the prediction results of different small sample processing algorithms, and Table 2 shows the various indicators of the prediction results of different cluster prediction strategies;

[0095] Table 1

[0096]

[0097] Table 2

[0098]

[0099] Embodiment 2:

[0100] The present invention also proposes a photovoltaic cluster short-term power prediction system 200 for sandstorm weather, such as Figure 5 As shown, including:

[0101] The data processing unit 201 is used to obtain weather data and power data under sandstorm weather in the photovoltaic area, pre-process the weather data and power data to divide the weather data and power data into sandstorm scenes, and expand small sample data in the weather data and power data to obtain a training set;

[0102] The prediction unit 202 is used to generate an adaptive combined prediction model based on the training set, input the test set into the adaptive combined prediction model, perform short-term photovoltaic power prediction, and obtain the short-term photovoltaic predicted power;

[0103] The output unit 203 is used to predict the short-term predicted power of the photovoltaic cluster based on the short-term predicted power of the photovoltaics based on the statistical upscaling algorithm.

[0104] Among them, weather data includes: numerical weather forecast data and sandstorm data.

[0105] The weather data and the power data are preprocessed to divide the weather data and the power data into dust scenes, including:

[0106] Determine the start and end time of the sandstorm event according to the weather data, calculate the influence of sandstorm concentration on power generation according to the start and end time of the sandstorm event, and divide the sandstorm weather scenes based on the influence of sandstorm concentration on power generation;

[0107] The sandstorm weather scene division is performed;

[0108] Setting a dust concentration threshold, and determining whether the dust will cause a loss in photovoltaic output based on the dust concentration threshold;

[0109] If, starting from a certain moment, the dust concentration increases and exceeds the dust concentration threshold within a preset time period, it is determined to be a dust-affected scene;

[0110] If the dust concentration begins to be less than the dust concentration threshold at a certain moment and does not rebound within a preset time period, it is determined as a scene without dust influence.

[0111] Among them, the small sample data in the weather data and power data are expanded to obtain the training set, including:

[0112] The data in the weather data and power data under dust scenes are taken as small samples, and the small samples are expanded using a time series generative adversarial network algorithm. The isolated forest algorithm is used to remove outliers from the expanded small samples to generate expanded sample data. The expanded sample data and the remaining data in the weather data and power data excluding the small sample data are constructed as a training set.

[0113] Among them, the pre-built adaptive combined prediction model includes: a pre-established photovoltaic power prediction benchmark model and a pre-established photovoltaic power prediction correction model under extreme sandstorm weather.

[0114] The test set is input into the adaptive combined prediction model to perform short-term photovoltaic power prediction, and the short-term photovoltaic power prediction is obtained, including:

[0115] The test set is divided into data affected by sandstorms and normal weather data according to the sandstorm scenarios, and marked. The normal weather data is predicted using the photovoltaic power prediction benchmark model of the adaptive combined prediction model to obtain a first prediction result. The data affected by sandstorms is predicted using the photovoltaic power prediction correction model of the adaptive combined prediction model to obtain a second prediction result. According to the mark, the first prediction result is combined with the second prediction result using a self-use algorithm to predict the short-term predicted power of the photovoltaic site.

[0116] Among them, based on the statistical upscaling algorithm, the short-term predicted power of the photovoltaic cluster is predicted according to the short-term predicted power of the photovoltaic site, including:

[0117] By sorting points to identify clustering structures, the OPTICS algorithm and the clustering K-Medoids algorithm are used to divide the photovoltaic cluster area into different sub-areas, and benchmark sites are selected for each sub-area. The output power of the benchmark site and the photovoltaic cluster area is calculated, and the average absolute error and correlation coefficient after normalization are calculated. Based on the average absolute error and correlation coefficient, the weight coefficient of the benchmark site is calculated;

[0118] Different weights are assigned to photovoltaics in different sub-regions, and based on the statistical upscaling algorithm, the short-term predicted power of the photovoltaic cluster area is predicted according to the short-term predicted power corresponding to photovoltaics with different weights.

[0119] The present invention can effectively predict the photovoltaic power under extreme weather conditions in the short term by building a model and a statistical upscaling algorithm.

[0120] Embodiment 3:

[0121] Based on the same inventive concept, the present invention also provides a computer device, which includes a processor and a memory, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute 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 (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, and is specifically suitable for loading and executing one or more instructions in the computer storage medium to implement the corresponding method flow or corresponding functions, so as to implement the steps of the method in the above embodiment.

[0122] Embodiment 4:

[0123] Based on the same inventive concept, the present invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device for storing programs and data. It is understandable that the computer-readable storage medium here can include both a built-in storage medium in a computer device and an extended storage medium supported by the computer device. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by a processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the steps of the method in the above embodiment.

[0124] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes. The schemes in the embodiments of the present invention may be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.

[0125] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0126] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0127] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0128] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0129] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for short-term power prediction of photovoltaic clusters in sandstorm weather, characterized in that: include: Acquire weather data and power data under sandstorm weather in the photovoltaic area, pre-process the weather data and power data to divide the weather data and power data into sandstorm scenes, and expand small sample data in the weather data and power data to obtain a training set; Based on the training set, an adaptive combined prediction model is generated, and the test set is input into the adaptive combined prediction model to perform short-term photovoltaic power prediction to obtain short-term photovoltaic predicted power; Based on the statistical upscaling algorithm, the short-term predicted power of the photovoltaic cluster is predicted according to the short-term predicted power of photovoltaics.

2. The power prediction method according to claim 1, characterized in that: The weather data include: numerical weather forecast data and sandstorm data.

3. The power prediction method according to claim 1, characterized in that: The preprocessing of the weather data and the power data to divide the weather data and the power data into dust scenes includes: Determine the start and end time of the sandstorm event according to the weather data, calculate the influence of sandstorm concentration on power generation according to the start and end time of the sandstorm event, and divide the sandstorm weather scenes based on the influence of sandstorm concentration on power generation; The sandstorm weather scene division is performed; Setting a dust concentration threshold, and determining whether the dust will cause a loss in photovoltaic output based on the dust concentration threshold; If, starting from a certain moment, the dust concentration increases and exceeds the dust concentration threshold within a preset time period, it is determined to be a dust-affected scene; If the dust concentration begins to be less than the dust concentration threshold at a certain moment and does not rebound within a preset time period, it is determined as a scene without dust influence.

4. The power prediction method according to claim 1, characterized in that: The small sample data in the weather data and power data are expanded to obtain a training set, including: The data in the weather data and power data under dust scenes are taken as small samples, and the small samples are expanded using a time series generative adversarial network algorithm. The isolated forest algorithm is used to remove outliers from the expanded small samples to generate expanded sample data. The expanded sample data and the remaining data in the weather data and power data excluding the small sample data are constructed as a training set.

5. The power prediction method according to claim 1, characterized in that: The pre-built adaptive combined prediction model includes: a pre-established photovoltaic power prediction benchmark model and a pre-established photovoltaic power prediction correction model under extreme sandstorm weather conditions.

6. The power prediction method according to claim 1, characterized in that: The test set is input into the adaptive combined prediction model to perform short-term photovoltaic power prediction to obtain the short-term photovoltaic predicted power, including: The test set is divided into data affected by sandstorms and normal weather data according to the sandstorm scenarios, and marked. The normal weather data is predicted using the photovoltaic power prediction benchmark model of the adaptive combined prediction model to obtain a first prediction result. The data affected by sandstorms is predicted using the photovoltaic power prediction correction model of the adaptive combined prediction model to obtain a second prediction result. According to the mark, the first prediction result is combined with the second prediction result using a self-use algorithm to predict the short-term predicted power of the photovoltaic site.

7. The power prediction method according to claim 1, characterized in that: The statistical upscaling algorithm is used to predict the short-term predicted power of the photovoltaic cluster according to the short-term predicted power of the photovoltaic site, including: By sorting points to identify clustering structures, the OPTICS algorithm and the clustering K-Medoids algorithm are used to divide the photovoltaic cluster area into different sub-areas, and benchmark sites are selected for each sub-area. The output power of the benchmark site and the photovoltaic cluster area is calculated, and the average absolute error and correlation coefficient after normalization are calculated. Based on the average absolute error and correlation coefficient, the weight coefficient of the benchmark site is calculated; Different weights are assigned to photovoltaics in different sub-regions, and based on the statistical upscaling algorithm, the short-term predicted power of the photovoltaic cluster area is predicted according to the short-term predicted power corresponding to photovoltaics with different weights.

8. A photovoltaic cluster short-term power prediction system for sandstorm weather, characterized in that: include: A data processing unit is used to obtain weather data and power data under sandstorm weather in the photovoltaic area, pre-process the weather data and power data to divide the weather data and power data into sandstorm scenes, and expand small sample data in the weather data and power data to obtain a training set; A prediction unit, used to generate an adaptive combined prediction model based on the training set, input the test set into the adaptive combined prediction model, perform short-term photovoltaic power prediction, and obtain the short-term photovoltaic predicted power; The output unit is used to predict the short-term predicted power of the photovoltaic cluster based on the short-term predicted power of the photovoltaic based on the statistical upscaling algorithm.

9. The power prediction system according to claim 8, characterized in that: The weather data include: numerical weather forecast data and sandstorm data.

10. The power prediction system according to claim 8, characterized in that: The preprocessing of the weather data and the power data to divide the weather data and the power data into dust scenes includes: Determine the start and end time of the sandstorm event according to the weather data, calculate the influence of sandstorm concentration on power generation according to the start and end time of the sandstorm event, and divide the sandstorm weather scenes based on the influence of sandstorm concentration on power generation; The sandstorm weather scene division is performed; Setting a dust concentration threshold, and determining whether the dust will cause a loss in photovoltaic output based on the dust concentration threshold; If, starting from a certain moment, the dust concentration increases and exceeds the dust concentration threshold within a preset time period, it is determined to be a dust-affected scene; If the dust concentration begins to be less than the dust concentration threshold at a certain moment and does not rebound within a preset time period, it is determined as a scene without dust influence.

11. The power prediction system according to claim 8, characterized in that: The small sample data in the weather data and power data are expanded to obtain a training set, including: The data in the weather data and power data under dust scenes are taken as small samples, and the small samples are expanded using a time series generative adversarial network algorithm. The isolated forest algorithm is used to remove outliers from the expanded small samples to generate expanded sample data. The expanded sample data and the remaining data in the weather data and power data excluding the small sample data are constructed as a training set.

12. The power prediction system according to claim 8, characterized in that: The pre-built adaptive combined prediction model includes: a pre-established photovoltaic power prediction benchmark model and a pre-established photovoltaic power prediction correction model under extreme sandstorm weather conditions.

13. The power prediction system according to claim 8, characterized in that: The test set is input into the adaptive combined prediction model to perform short-term photovoltaic power prediction to obtain the short-term photovoltaic predicted power, including: The test set is divided into data affected by sandstorms and normal weather data according to the sandstorm scenarios, and marked. The normal weather data is predicted using the photovoltaic power prediction benchmark model of the adaptive combined prediction model to obtain a first prediction result. The data affected by sandstorms is predicted using the photovoltaic power prediction correction model of the adaptive combined prediction model to obtain a second prediction result. According to the mark, the first prediction result is combined with the second prediction result using a self-use algorithm to predict the short-term predicted power of the photovoltaic site.

14. The power prediction system according to claim 8, characterized in that: The statistical upscaling algorithm is used to predict the short-term predicted power of the photovoltaic cluster according to the short-term predicted power of the photovoltaic site, including: By sorting points to identify clustering structures, the OPTICS algorithm and the clustering K-Medoids algorithm are used to divide the photovoltaic cluster area into different sub-areas, and benchmark sites are selected for each sub-area. The output power of the benchmark site and the photovoltaic cluster area is calculated, and the average absolute error and correlation coefficient after normalization are calculated. Based on the average absolute error and correlation coefficient, the weight coefficient of the benchmark site is calculated; Different weights are assigned to photovoltaics in different sub-regions, and based on the statistical upscaling algorithm, the short-term predicted power of the photovoltaic cluster area is predicted according to the short-term predicted power corresponding to photovoltaics with different weights.

15. A computer device, characterized in that: include: one or more processors; a processor for executing one or more programs; When the one or more programs are executed by the one or more processors, the method according to any one of claims 1 to 7 is implemented.

16. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed, the method according to any one of claims 1 to 7 is implemented.

Images