Photovoltaic power generation power model construction and prediction method, device, equipment and medium

Abstract

The invention provides a photovoltaic power generation power model construction and prediction method and device, equipment and a medium, and relates to the technical field of photovoltaic power generation power prediction, and the model construction method comprises the steps: obtaining an original data set; wherein each original data subset in the original data set comprises historical generated power data, historical cloud layer images and historical meteorological data of various types of meteorological parameters; generating a first training data set according to the original data set; and performing model training on a pre-constructed initial photovoltaic power generation power prediction model by using the first training data set to obtain a target photovoltaic power generation power prediction model. The target photovoltaic power generation power prediction model obtained through training is used for photovoltaic power generation power prediction, and the problem that an existing photovoltaic power generation power prediction method does not consider the influence of cloud layer changes on the photovoltaic power generation power, and consequently the accuracy of model prediction is low is solved.

Description

Technical Field

[0001] The present application relates to the technical field of photovoltaic power generation prediction, and specifically to a photovoltaic power generation prediction model construction method, a photovoltaic power generation prediction method, a photovoltaic power generation prediction model construction device, a photovoltaic power generation prediction device, an electronic device and a machine-readable storage medium. Background Art

[0002] The power generation of a photovoltaic power station (hereinafter referred to as photovoltaic power generation) is affected by many factors. Improving the accuracy of photovoltaic power generation prediction will help stabilize the operation of the power system.

[0003] In related technologies, the prediction of photovoltaic power generation is usually based on statistical methods. Specifically, support vector machines, regression analysis and other models are used to build prediction models, and numerical historical meteorological data (such as the numerical values ​​corresponding to solar radiation intensity, illumination time, temperature, wind speed and direction, and humidity) and historical power generation data are used to train the model. By associating numerical meteorological data with power generation data, the trained model can be used to predict the future power generation of photovoltaic power stations.

[0004] However, when clouds pass through the area of ​​PV power stations, they may block the PV panels, resulting in less sunlight received by the panels. This blocking phenomenon will cause sudden fluctuations in PV power generation, resulting in a large deviation between the actual power generation of the PV power station and the power generation predicted by the prediction model, affecting the load balance of the power grid. Summary of the invention

[0005] The purpose of the embodiments of the present application is to provide a photovoltaic power generation model construction and prediction method, device, equipment and medium to solve the problem that the existing photovoltaic power generation power prediction method does not consider the impact of cloud changes on photovoltaic power generation, resulting in low accuracy of model prediction.

[0006] In order to achieve the above-mentioned purpose, the first aspect of the present application provides a method for constructing a photovoltaic power generation power prediction model, the method comprising:

[0007] Acquire an original data set; wherein each original data subset in the original data set includes historical power generation data, historical cloud layer images, and historical meteorological data of multiple types of meteorological parameters;

[0008] Generate a first training data set according to the original data set; wherein each first training data subset in the first training data set corresponds to an original data subset;

[0009] The first training data set is used to perform model training on the pre-built initial photovoltaic power generation prediction model to obtain a target photovoltaic power generation prediction model.

[0010] In the embodiment of the present application, generating a first training data set according to the original data set includes:

[0011] For each historical cloud layer image, annotate the historical cloud layer image to obtain an annotation result of the historical cloud layer image; wherein the annotation result of the historical cloud layer image includes historical cloud amount data corresponding to the historical cloud layer image;

[0012] Generate a cloud amount data set based on each historical cloud layer image; wherein the historical cloud amount data corresponding to each historical cloud layer image is the first cloud amount data in the cloud amount data set;

[0013] Based on the cloud cover dataset, determine the mean and variance corresponding to the cloud layer image;

[0014] Determine normalized data corresponding to each first cloud amount data in the cloud amount data set according to the mean and variance corresponding to the cloud layer image;

[0015] The first training data set includes normalized data corresponding to each first cloud amount data in the cloud amount data set.

[0016] In an embodiment of the present application, the method further includes:

[0017] The pre-built initial image quantization model is trained using the annotation results of each historical cloud image to obtain the target image quantization model.

[0018] In the embodiment of the present application, generating a first training data set according to the original data set includes:

[0019] For each type of meteorological parameter, based on each historical meteorological data of the meteorological parameter, a first meteorological data set corresponding to the meteorological parameter is generated; wherein each historical meteorological data of the meteorological parameter is the first meteorological data in the first meteorological data set corresponding to the meteorological parameter;

[0020] For each type of meteorological parameter, determine the mean and variance corresponding to the type of meteorological parameter according to the first meteorological data set corresponding to the type of meteorological parameter;

[0021] For each type of meteorological parameter, determine the normalized data corresponding to each of the first meteorological data in the first meteorological data set corresponding to the meteorological parameter according to the mean and variance corresponding to the meteorological parameter;

[0022] For each type of meteorological parameter, the first training data set includes normalized data corresponding to each first meteorological data in the first meteorological data set corresponding to the meteorological parameter of this type.

[0023] In an embodiment of the present application, the method further includes:

[0024] The second original data is obtained according to the first preset time interval, and the following steps are performed: new cloud amount data corresponding to the new cloud layer image in the second original data is determined, and according to the new cloud amount data, whether the current target photovoltaic power generation prediction model meets the preset incremental training conditions is determined; if so, a second training data set is generated according to the original data set and each second original data, and the current target photovoltaic power generation prediction model is incrementally trained using the second training data set to obtain a new target photovoltaic power generation prediction model.

[0025] A second aspect of the present application provides a photovoltaic power generation power prediction method, the method comprising:

[0026] Acquire target data; wherein the target data includes target cloud images corresponding to a plurality of consecutive target acquisition time points and target meteorological data of multiple types of meteorological parameters;

[0027] Based on the target photovoltaic power generation prediction model described in the first aspect, a photovoltaic power generation prediction result corresponding to the target data is determined.

[0028] The third aspect of the present application provides a photovoltaic power generation power prediction model construction device, the device comprising:

[0029] The original data set acquisition module is used to acquire the original data set; wherein each original data subset in the original data set includes historical power generation data, historical cloud layer images and historical meteorological data of multiple types of meteorological parameters;

[0030] A training data set generation module, used to generate a first training data set according to the original data set; wherein each first training data subset in the first training data set corresponds to an original data subset;

[0031] The first prediction model generation module is used to perform model training on a pre-built initial photovoltaic power generation prediction model using a first training data set to obtain a target photovoltaic power generation prediction model.

[0032] A fourth aspect of the present application provides a photovoltaic power generation prediction device, the photovoltaic power generation prediction device comprising:

[0033] A target data acquisition module is used to acquire target data; wherein the target data includes target cloud images corresponding to a plurality of consecutive target acquisition time points and target meteorological data of multiple types of meteorological parameters;

[0034] The photovoltaic power generation prediction module is used to determine the photovoltaic power generation prediction result corresponding to the target data based on the target photovoltaic power generation prediction model described in the first aspect.

[0035] The fifth aspect of the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the photovoltaic power generation prediction model construction method described in the first aspect and / or the photovoltaic power generation prediction method described in the second aspect are implemented.

[0036] In a sixth aspect, the present application provides a machine-readable storage medium having instructions stored thereon, which, when executed by a processor, configures the processor to execute the photovoltaic power generation prediction model construction method described in the first aspect, and / or the photovoltaic power generation prediction method described in the second aspect.

[0037] The photovoltaic power generation model construction and prediction method, device, equipment and medium provided in the present application obtain historical power generation data, historical cloud images and historical meteorological data of multiple types of meteorological parameters, and generate a first training data set based on the acquired data. The initial photovoltaic power generation prediction model is trained through the first training data set, which can improve the prediction accuracy of the trained target photovoltaic power generation prediction model, and solve the problem that the existing photovoltaic power generation prediction method does not consider the impact of cloud changes on photovoltaic power generation, resulting in low model prediction accuracy. After the target data (including cloud images and meteorological data) is subsequently obtained, the photovoltaic power generation prediction result corresponding to the target data can be predicted in real time based on the target photovoltaic power generation prediction model.

[0038] Other features and advantages of the embodiments of the present application will be described in detail in the subsequent specific implementation section. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present application, but do not constitute a limitation on the embodiments of the present application. In the accompanying drawings:

[0040] Figure 1 The schematic diagram of the process of constructing a photovoltaic power prediction model according to an embodiment of the present application is shown;

[0041] Figure 2 A schematic diagram schematically shows a normalized image according to an embodiment of the present application;

[0042] Figure 3 The schematic diagram of the process of the photovoltaic power generation prediction method according to the embodiment of the present application is shown schematically;

[0043] Figure 4 A schematic diagram of a photovoltaic power generation prediction model construction device according to an embodiment of the present application is shown;

[0044] Figure 5 A first structural block diagram of a photovoltaic power generation prediction device according to an embodiment of the present application is schematically shown;

[0045] Figure 6 A second structural block diagram of the photovoltaic power generation prediction device according to an embodiment of the present application is schematically shown;

[0046] Figure 7 The internal structure diagram of the computer device according to the embodiment of the present application is schematically shown.

[0047] Description of Reference Numerals

[0048] A01-processor; A02-network interface; A03-internal memory; A04-display screen; A05-input device; A06-non-volatile storage medium; B01-operating system; B02-computer program. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. It should be understood that the specific implementation methods described herein are only used to illustrate and explain the embodiments of the present application, and are not used to limit the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0050] It should be noted that if the embodiments of the present application involve directional indications (such as up, down, left, right, front, back, etc.), such directional indications are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0051] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present application, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or suggesting their relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by this application.

[0052] In view of the fact that the existing photovoltaic power generation prediction model construction method in the related art does not take into account the impact of cloud changes on photovoltaic power generation, resulting in the problem of low accuracy of model prediction, the present application provides a photovoltaic power generation model construction and prediction method, device, equipment and medium. The photovoltaic power generation model construction and prediction method, device, equipment and medium provided in the present application are described in detail below in conjunction with the accompanying drawings through specific examples and implementation methods.

[0053] Figure 1 The schematic diagram of the process of constructing the photovoltaic power prediction model of the embodiment of the present application is shown schematically. Figure 1 As shown, in one embodiment of the present application, a method for constructing a photovoltaic power generation prediction model is provided, and the method for constructing a photovoltaic power generation prediction model may include the following steps.

[0054] Step A200, obtaining the original data set.

[0055] The original data set includes multiple original data subsets, each of which includes historical power generation data, historical cloud images, and historical meteorological data of multiple types of meteorological parameters.

[0056] Specifically, the historical cloud layer image is an image of the cloud layer above the photovoltaic power station during its historical operation. In this embodiment, the historical cloud layer image can be obtained by a sky camera or other image acquisition equipment, and this application does not specifically limit the method for obtaining the historical cloud layer image.

[0057] In this embodiment, the various meteorological parameters may be solar radiation intensity, illumination time, temperature, wind speed and direction, humidity, etc. The historical meteorological data of various meteorological parameters are the values ​​of such meteorological parameters during the historical operation of the photovoltaic power station.

[0058] It should be noted that, in this embodiment, for each raw data subset, the raw data subset includes first raw data corresponding to each of a plurality of consecutive first historical collection time points, and second raw data corresponding to each of a plurality of second historical collection time points after the plurality of first historical collection time points (i.e., second raw data within a future period of time), wherein each first raw data includes historical cloud layer images and historical meteorological data of multiple types of meteorological parameters, and each second raw data includes historical power generation data. It can be understood that for the same time point, it may be either the first historical collection time point or the second historical collection time point.

[0059] Step A400: Generate a first training data set based on the original data set.

[0060] Wherein, each first training data subset in the first training data set corresponds to an original data subset.

[0061] In this embodiment, the first training data subset includes normalized data corresponding to each historical power generation data in its corresponding original data subset, normalized data corresponding to each historical cloud layer image, and normalized data corresponding to each historical meteorological data.

[0062] As a specific implementation, step A400 may include the following steps:

[0063] Step A411 : for each historical cloud layer image, annotate the historical cloud layer image to obtain an annotation result of the historical cloud layer image.

[0064] Each historical cloud layer image can be annotated by a computer program (such as an existing annotation tool) or manually, and the annotation result of the historical cloud layer image includes the historical cloud cover data corresponding to the historical cloud layer image and the minimum bounding box containing the largest cloud layer in the historical cloud layer image. It is worth mentioning that the historical cloud cover data corresponding to the historical cloud layer image is a specific value, and its value range is between 0 and 1.

[0065] Step A412, generating a cloud cover dataset based on each historical cloud layer image.

[0066] Due to reasons such as equipment failure, there may be some first historical collection time points without corresponding historical cloud layer images, that is, the cloud layer images at these first historical collection time points are not collected. Based on this, in this embodiment, the step A412 specifically includes:

[0067] For each first historical collection time point, determine whether there is a corresponding historical cloud layer image for the first historical collection time point. If so, determine the first cloud amount data corresponding to the first historical collection time point as the historical cloud amount data corresponding to the historical cloud layer image corresponding to the first historical collection time point; if not, determine the first cloud amount data corresponding to the first historical collection time point as the preset cloud amount data.

[0068] The preset cloud amount data may be selected as 0. Of course, it is understandable that the preset cloud amount data may also be set to other values ​​between 0 and 1, and the setting of the preset cloud amount data may be adjusted according to actual conditions.

[0069] In this embodiment, the cloud amount data set includes the first cloud amount data corresponding to each first historical collection time point, that is, the historical cloud amount data corresponding to each historical cloud layer image is the first cloud amount data in the cloud amount data set.

[0070] Step A413, determining the mean and variance corresponding to the cloud layer image according to the cloud cover data set.

[0071] In this embodiment, in step A413, the mean corresponding to the cloud layer image is determined by a first formula according to each first cloud amount data in the cloud amount data set, and the variance corresponding to the cloud layer image is determined by a second formula.

[0072] Among them, the first formula is:

[0073]

[0074] In the above formula, μ represents the mean value corresponding to the cloud layer image, n represents the total number of the first cloud amount data in the cloud amount dataset, and x i Represents the i-th first cloud cover data in the cloud cover dataset.

[0075] The second formula is:

[0076]

[0077] In the above formula, σ represents the variance corresponding to the cloud image.

[0078] Step A414, determining normalized data corresponding to each first cloud amount data in the cloud amount data set according to the mean and variance corresponding to the cloud layer image.

[0079] In this embodiment, in step A414, for each first cloud cover data, the normalized data corresponding to the first cloud cover data is determined by the third formula to remove the dimension of the data.

[0080] Among them, the third formula is:

[0081]

[0082] In the above formula, x i ′ Indicates the first cloud cover data x i The corresponding normalized data.

[0083] In this embodiment, the first training data set includes normalized data corresponding to each first cloud amount data in the cloud amount data set. Specifically, each first training data subset includes normalized data corresponding to the first cloud amount data corresponding to each first historical collection time point in its corresponding original data subset.

[0084] In this specific implementation, the step A400 may further include the following steps:

[0085] Step A421: Generate a first power generation data set based on various historical power generation data.

[0086] Among them, each piece of historical power generation data is the first power generation data in the first power generation data set.

[0087] Step A422: Determine the mean and variance corresponding to the photovoltaic power generation according to the first power generation data set.

[0088] Step A423: Determine normalized data corresponding to each first power generation data in the first power generation data set according to the mean and variance corresponding to the photovoltaic power generation.

[0089] The first training data set includes normalized data corresponding to each first power generation data in the first power generation data set. Specifically, each first training data subset includes normalized data corresponding to the first power generation data corresponding to each second historical acquisition time point in its corresponding original data subset.

[0090] In this specific implementation, for each type of meteorological parameter, step A400 may further include the following steps:

[0091] Step A431, generating a first meteorological data set corresponding to the meteorological parameter of this type based on the historical meteorological data of the meteorological parameter of this type.

[0092] For each type of meteorological parameter, each historical meteorological data of the meteorological parameter is the first meteorological data in the first meteorological data set corresponding to the meteorological parameter.

[0093] Step A432: Determine the mean and variance corresponding to the meteorological parameter of this type according to the first meteorological data set corresponding to the meteorological parameter of this type.

[0094] Step A433, determining normalized data corresponding to each first meteorological data in the first meteorological data set corresponding to the meteorological parameter of this type according to the mean and variance corresponding to the meteorological parameter of this type.

[0095] For each type of meteorological parameter, the first training data set includes normalized data corresponding to each first meteorological data in the first meteorological data set corresponding to the meteorological parameter. Specifically, each first training data subset includes normalized data corresponding to all first meteorological data corresponding to each first historical collection time point in its corresponding original data subset.

[0096] It should be noted that, for each type of meteorological parameter, when historical meteorological data at a certain historical collection time point is not collected, the first meteorological data corresponding to the first historical collection time point is determined as the preset meteorological data corresponding to this type of meteorological parameter.

[0097] In this specific implementation, the method of generating the first power generation data set and the method of generating the first meteorological data set corresponding to each type of meteorological parameter are similar to the method of generating the cloud cover data set, and will not be repeated here; the method of determining the mean and variance corresponding to the photovoltaic power generation power and the method of determining the mean and variance corresponding to each type of meteorological parameter are similar to the method of determining the mean and variance corresponding to the cloud image, and will not be repeated here; the method of determining the normalized data corresponding to each first power generation data and the method of determining the normalized data corresponding to each first meteorological data are similar to the method of determining the normalized data corresponding to each first cloud cover data, and will not be repeated here.

[0098] Step A600: Use the first training data set to perform model training on the pre-built initial photovoltaic power prediction model to obtain a target photovoltaic power prediction model.

[0099] In this embodiment, the initial photovoltaic power generation prediction model adopts a long short-term memory network (Long Short-Term Memory, LSTM) model. Specifically, the LSTM model can effectively solve the gradient vanishing and gradient explosion problems faced by the traditional recurrent neural network (Recurrent Neural Network, RNN) model when processing long sequences by introducing a gating mechanism, thereby better capturing long-term dependencies. In one example, the initial photovoltaic power generation prediction model adopts an LSTM model with a hidden layer size of 128 and a stacking layer number of 2 to improve the model training speed and model prediction accuracy.

[0100] It is understandable that since the target photovoltaic power generation prediction model is obtained by training the initial photovoltaic power generation prediction model, the model structures of the two are the same. Of course, it is understandable that the initial photovoltaic power generation prediction model can also adopt other network structures suitable for processing time series data prediction, and this application does not specifically limit the structure of the initial photovoltaic power generation prediction model.

[0101] In this embodiment, when the initial photovoltaic power generation prediction model is trained using the first training data set, the normalized data corresponding to the historical cloud images and the normalized data corresponding to the historical meteorological data in the first training data subset are used as model inputs, and the normalized data corresponding to the historical power generation data in the first training data subset are used as model outputs, so as to achieve the trained target photovoltaic power generation prediction model, which can predict the photovoltaic power generation data within a certain period of time in the future based on the collected cloud images and meteorological data of various meteorological parameters.

[0102] Optionally, in this embodiment, after step A411, the photovoltaic power generation power prediction model construction method may further include the following steps:

[0103] Step A700, using the annotation results of each historical cloud layer image, the pre-built initial image quantization model is trained to obtain a target image quantization model.

[0104] In this embodiment, the initial image quantization model adopts the YOLO model (e.g., YOLOv5). It is understandable that since the target image quantization model is obtained by training the initial image quantization model, the model structures of the two are the same. Of course, it is understandable that the initial image quantization model can also adopt other model structures that can realize quantization of a certain / certain specific objects in the image, and this application does not specifically limit the initial image quantization model structure.

[0105] In this embodiment, for the trained target image quantization model, after any cloud cover image collected subsequently is input into the model, the target image quantization model performs the following operations:

[0106] First, the cloud cover image is normalized to obtain a normalized image corresponding to the cloud cover image. For example, the cloud cover image is normalized to an image with a size of 960×960 pixels.

[0107] Then, a minimum bounding box containing a maximum cloud layer in the normalized image and parameter values ​​of multiple parameters of the minimum bounding box are determined.

[0108] Among them, the parameters of the minimum bounding box include the coordinates of the center position of the largest cloud layer in the normalized image, the cloud layer height ratio, the cloud layer width ratio and the confidence score. It should be noted that the target image quantization model will first generate a large number of candidate bounding boxes during image processing. For each candidate bounding box, the target image quantization model will output the parameter value of the confidence score of the candidate bounding box, which reflects the probability of the existence of the cloud layer. Then, non-maximum suppression will be used to process each candidate bounding box, that is, only the box with the highest confidence score (that is, the minimum bounding box) is retained. In addition, the parameter values ​​of each parameter of the minimum bounding box are determined relative to the entire normalized image. For example, the cloud layer height ratio represents the ratio of the height of the minimum bounding box (that is, the cloud layer height) to the height of the entire normalized image, and the cloud layer width ratio represents the ratio of the width of the minimum bounding box (that is, the cloud layer width) to the width of the entire normalized image.

[0109] Finally, based on the normalized image and the parameter values ​​of the minimum bounding box, the cloud amount data corresponding to the cloud amount image is determined and output by the fourth formula.

[0110] Among them, the fourth formula is:

[0111]

[0112] In the above formula, s represents the cloud data corresponding to the cloud image, w represents the cloud width ratio value, h represents the cloud height ratio value, A and B represent the width and height values ​​of the normalized image respectively, and a and b represent the width and height values ​​of the minimum bounding box in the normalized image respectively.

[0113] It can be understood that the quantitative data corresponding to the cloud cover image is the ratio of the area of ​​the minimum bounding box in the normalized image corresponding to the cloud cover image to the total area of ​​the normalized image.

[0114] Figure 2 The schematic diagram of the normalized image of the embodiment of the present application is schematically shown. Figure 2 As shown, in one example, if the size of the normalized image of a cloud cover image is 500×400 pixels, and the width of the minimum bounding box in the normalized image is 400 pixels and the height is 250 pixels, then the quantitative data corresponding to the cloud cover image is 0.5.

[0115] This embodiment uses the annotation results of historical cloud images to perform model training, so as to efficiently and accurately obtain the cloud amount data corresponding to the cloud images in the future, thereby further improving the prediction efficiency and accuracy of photovoltaic power generation.

[0116] Optionally, in this embodiment, after step A700, the photovoltaic power generation power prediction model construction method may further include the following steps:

[0117] Step A800, obtaining new raw data according to a first preset time interval, and performing the following steps: determining new cloud cover data corresponding to a new cloud layer image in the new raw data, determining whether the current target photovoltaic power generation prediction model meets the preset incremental training conditions based on the new cloud cover data, and if so, generating a second training data set based on the original data set and each new raw data, and using the second training data set to perform incremental training on the current target photovoltaic power generation prediction model to obtain a new target photovoltaic power generation prediction model.

[0118] The new raw data includes a new cloud layer image corresponding to a single new acquisition time point, new power generation data, and new meteorological data of multiple types of meteorological parameters. In this embodiment, for a new cloud layer image, by inputting it into the target image quantization model, new cloud amount data corresponding to the new cloud layer image can be obtained.

[0119] In this embodiment, the first preset time interval can be selected as 5 minutes, that is, new raw data is collected every 5 minutes. Of course, it can be understood that the first preset time interval can also be adjusted according to actual conditions, and this application does not specifically limit the selection of the first preset time interval.

[0120] In this embodiment, the preset incremental training condition can be: the latest time difference is less than or equal to the second preset time interval corresponding to the new cloud cover data, wherein the latest time difference is the time difference between the time point of the last training of the target photovoltaic power generation prediction model and the time point of collecting new original data.

[0121] In one example, when the new cloud cover data y satisfies 0≤y<0.5, the corresponding second preset time interval is selected as 30 minutes; when the new cloud cover data y satisfies 0.5≤y<0.75, the corresponding second preset time interval is selected as 20 minutes; when the new cloud cover data y satisfies 0.75≤y≤1, the corresponding second preset time interval is selected as 15 minutes.

[0122] In this embodiment, the second training data set includes multiple second training data subsets, each first training data subset is a second training data subset included in the second training data set, and after acquiring new original data, a second training data subset is formed based on the new original data.

[0123] Among them, in step A800, when generating the second training data subset based on the new original data, it is necessary to determine the normalized data corresponding to each data in the new original data. When there is no new cloud layer image in the new original data or the new meteorological data lacking at least some meteorological parameters, the corresponding normalized data is still set to the corresponding preset value (such as preset cloud amount data, preset meteorological data). In this embodiment, the method for determining the normalized data corresponding to each data in the new original data is similar to the method for determining the normalized data corresponding to each data in the first original data / the second original data, and will not be repeated here.

[0124] It can be understood that the obtained new target photovoltaic power generation prediction model is the current target photovoltaic power generation prediction model.

[0125] In this embodiment, the method of incrementally training the current target photovoltaic power prediction model using the second training data set is similar to the method of training the initial photovoltaic power prediction model using the first training data set, and will not be repeated here.

[0126] This embodiment adjusts the frequency of incremental training of the target photovoltaic power prediction model according to the randomness and uncertainty of cloud obstruction, so that the model can adapt to the latest data changes more quickly and improve the accuracy and stability of photovoltaic power prediction when clouds are obstructed.

[0127] It is not difficult to see that the photovoltaic power generation prediction model construction method provided in this embodiment obtains historical power generation data, historical cloud images and historical meteorological data of multiple types of meteorological parameters, and generates a first training data set based on the acquired data. The initial photovoltaic power generation prediction model is trained through the first training data set, which can improve the prediction accuracy of the trained target photovoltaic power generation prediction model, and solves the problem that the existing photovoltaic power generation prediction method does not consider the impact of cloud changes on photovoltaic power generation, resulting in low model prediction accuracy.

[0128] Figure 1 FIG. 1 is a flow chart of a method for constructing a photovoltaic power generation prediction model in one embodiment. It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0129] Figure 3 The schematic diagram of the process flow of the photovoltaic power generation prediction method of the embodiment of the present application is shown schematically. Figure 3 As shown, in one embodiment of the present application, a photovoltaic power generation power prediction method is provided, and the photovoltaic power generation power prediction method may include the following steps.

[0130] Step B200, obtaining target data.

[0131] The target data includes target cloud images corresponding to a plurality of consecutive target acquisition time points and target meteorological data of multiple types of meteorological parameters. It can be understood that the target data may include a plurality of new raw data acquired based on the first preset time interval.

[0132] Step B400, based on the target photovoltaic power generation prediction model, determining the photovoltaic power generation prediction result corresponding to the target data.

[0133] In this embodiment, step B400 may include the following steps:

[0134] Step B410, determining normalized data corresponding to each target cloud layer image and each target meteorological data.

[0135] Wherein, the step B410 specifically includes:

[0136] For each target cloud layer image, the target cloud amount data corresponding to the target cloud layer image is determined by the target image quantization model, and based on the target cloud amount data corresponding to the target cloud layer image, the normalized data corresponding to the target cloud layer image is determined by the third formula.

[0137] Specifically, for each target cloud layer image, by substituting the target cloud amount data corresponding to the target cloud layer image into the third formula, the normalized data corresponding to the target cloud layer image can be obtained.

[0138] For each type of target meteorological data of a meteorological parameter, the normalized data corresponding to the target meteorological data of the meteorological parameter can be determined based on the mean and variance corresponding to the meteorological parameter. In this embodiment, the method for determining the normalized data corresponding to the target meteorological data is similar to the method for determining the normalized data corresponding to the target cloud layer image, and will not be described in detail here.

[0139] Step B420, inputting the normalized data corresponding to each target cloud layer image and each target meteorological data into the target photovoltaic power generation prediction model to obtain the photovoltaic power generation prediction result corresponding to the target data.

[0140] Specifically, after the normalized data corresponding to each target cloud image and each target meteorological data are input into the current target photovoltaic power generation prediction model, the target photovoltaic power generation prediction model will output the normalized data corresponding to the predicted photovoltaic power generation data in a period of time in the future. Based on the mean and variance corresponding to the photovoltaic power generation power, by denormalizing each normalized data, the photovoltaic power generation data in the future can be obtained, which is the photovoltaic power generation power prediction result corresponding to the target data.

[0141] In this embodiment, based on the current target photovoltaic power prediction model, according to the target cloud images corresponding to multiple consecutive target acquisition time points and the target meteorological data of multiple types of meteorological parameters, the photovoltaic power data in a future period after the multiple target acquisition time points can be predicted in real time with high prediction accuracy.

[0142] The photovoltaic power generation power prediction method provided in this embodiment can perform photovoltaic power generation power prediction according to actual needs. For example, the method can perform ultra-short-term photovoltaic power generation power prediction within the next hour, and can also perform short-term photovoltaic power generation power prediction within the next 72 hours. For the photovoltaic power generation power prediction result obtained, the photovoltaic power generation power prediction result and the target data for obtaining the photovoltaic power generation power prediction result can be reported to the dispatching agency in real time to optimize the operation and management of the power grid.

[0143] Figure 4 The structure block diagram of the photovoltaic power generation prediction model construction device of the embodiment of the present application is schematically shown. Figure 4 As shown, in one embodiment of the present application, a photovoltaic power generation power prediction model construction device is provided, and the photovoltaic power generation power prediction model construction device may include:

[0144] The original data set acquisition module is used to acquire the original data set; wherein each original data subset in the original data set includes historical power generation data, historical cloud layer images and historical meteorological data of multiple types of meteorological parameters.

[0145] The training data set generation module is used to generate a first training data set according to the original data set; wherein each first training data subset in the first training data set corresponds to an original data subset.

[0146] The first prediction model generation module is used to perform model training on a pre-built initial photovoltaic power generation prediction model using a first training data set to obtain a target photovoltaic power generation prediction model.

[0147] In this embodiment, the training data set generation module may include:

[0148] The image annotation unit is used to annotate each historical cloud layer image to obtain the annotation result of the historical cloud layer image; wherein the annotation result of the historical cloud layer image includes the historical cloud amount data corresponding to the historical cloud layer image.

[0149] The cloud amount data set generation unit is used to generate a cloud amount data set based on each historical cloud layer image; wherein the historical cloud amount data corresponding to each historical cloud layer image is the first cloud amount data in the cloud amount data set.

[0150] The first related data determination unit is used to determine the mean and variance corresponding to the cloud layer image according to the cloud cover data set.

[0151] The first normalized data determining unit is used to determine the normalized data corresponding to each first cloud amount data in the cloud amount data set according to the mean and variance corresponding to the cloud layer image.

[0152] The first training data set includes normalized data corresponding to each first cloud amount data in the cloud amount data set.

[0153] In this embodiment, the photovoltaic power generation power prediction model building device may also include:

[0154] The quantization model generation module is used to use the annotation results of each historical cloud layer image to train the pre-built initial image quantization model to obtain the target image quantization model.

[0155] In this embodiment, the training data set generation module may further include:

[0156] The meteorological data set generation unit is used to generate a first meteorological data set corresponding to each type of meteorological parameter based on the various historical meteorological data of the meteorological parameter; wherein the various historical meteorological data of the meteorological parameter are the first meteorological data in the first meteorological data set corresponding to the meteorological parameter.

[0157] The second related data determination unit is used to determine, for each type of meteorological parameter, a mean value and a variance corresponding to the type of meteorological parameter according to the first meteorological data set corresponding to the type of meteorological parameter.

[0158] The second normalized data determination unit is used to determine, for each type of meteorological parameter, the normalized data corresponding to each of the first meteorological data in the first meteorological data set corresponding to the type of meteorological parameter according to the mean and variance corresponding to the type of meteorological parameter.

[0159] For each type of meteorological parameter, the first training data set includes normalized data corresponding to each first meteorological data in the first meteorological data set corresponding to the meteorological parameter of this type.

[0160] In this embodiment, the photovoltaic power generation power prediction model building device may also include:

[0161] The second prediction model generation module is used to obtain the second original data according to the first preset time interval, and perform the following operations: determine the new cloud amount data corresponding to the new cloud layer image in the second original data, and determine whether the current target photovoltaic power generation prediction model meets the preset incremental training conditions based on the new cloud amount data; if so, generate a second training data set based on the original data set and each second original data, and use the second training data set to perform incremental training on the current target photovoltaic power generation prediction model to obtain a new target photovoltaic power generation prediction model.

[0162] In this embodiment, the photovoltaic power generation prediction model construction device includes a processor and a memory. The original data set acquisition module, the training data set generation module and the first prediction model generation module are all stored in the memory as program units, and the processor executes the above program modules stored in the memory to implement corresponding functions.

[0163] The processor includes a kernel, and the kernel calls the corresponding program unit from the memory. One or more kernels can be set, and the photovoltaic power generation power prediction model construction method is implemented by adjusting kernel parameters.

[0164] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0165] Figure 5 The first structural block diagram of the photovoltaic power prediction device according to the embodiment of the present application is schematically shown. Figure 5 As shown, in one embodiment of the present application, a photovoltaic power generation prediction device is provided, and the photovoltaic power generation prediction device may include:

[0166] The target data acquisition module is used to acquire target data; wherein the target data includes target cloud layer images corresponding to multiple consecutive target acquisition time points and target meteorological data of multiple types of meteorological parameters.

[0167] The photovoltaic power generation prediction module is used to determine the photovoltaic power generation prediction result corresponding to the target data based on the target photovoltaic power generation prediction model.

[0168] In this embodiment, the photovoltaic power generation prediction module may include:

[0169] The image processing unit is used to determine, for each target cloud layer image, target cloud amount data corresponding to the target cloud layer image through a target image quantization model, and determine normalized data corresponding to the target cloud layer image through a third formula based on the target cloud amount data corresponding to the target cloud layer image.

[0170] The data processing unit is used to determine the normalized data corresponding to the target meteorological data of each type of meteorological parameter based on the mean and variance corresponding to the meteorological parameter of this type.

[0171] The prediction unit is used to input the normalized data corresponding to each target cloud layer image and each target meteorological data into the target photovoltaic power generation prediction model to obtain the photovoltaic power generation prediction result corresponding to the target data.

[0172] In this embodiment, the photovoltaic power generation prediction device includes a processor and a memory. The various functional modules included in the photovoltaic power generation prediction device (such as the target data acquisition module and the photovoltaic power generation prediction module) are stored in the memory as program units, and the processor executes the above program modules stored in the memory to implement corresponding functions.

[0173] In this embodiment, the structure and function of the processor and memory included in the photovoltaic power generation prediction device are similar to the structure and function of the processor and memory included in the photovoltaic power generation prediction model construction device, and will not be repeated here.

[0174] Figure 6 The second structural block diagram of the photovoltaic power prediction device according to the embodiment of the present application is schematically shown. Figure 6 As shown, this embodiment encapsulates the various functional modules included in the photovoltaic power prediction device in a low-power CPU and supplies power through a built-in battery, thereby forming a set of lightweight, portable and movable photovoltaic power prediction equipment. There is no need for additional dedicated workstations and external power supplies for continuous power supply, which solves the problems of limited use scenarios and high use costs of existing photovoltaic power prediction devices. In addition, the photovoltaic power prediction device provided by this embodiment has the advantages of stable performance and high prediction accuracy, which can meet the needs of long-term continuous operation.

[0175] An embodiment of the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the above-mentioned photovoltaic power generation prediction model construction method and / or the above-mentioned photovoltaic power generation prediction method are implemented.

[0176] An embodiment of the present application provides a machine-readable storage medium having instructions stored thereon, which, when executed by a processor, configures the processor to execute the above-mentioned photovoltaic power generation prediction model construction method and / or the above-mentioned photovoltaic power generation prediction method.

[0177] The present application provides a computer program product, which, when executed on a data processing device, is suitable for executing a program for initializing the following method steps:

[0178] Acquire an original data set; wherein each original data subset in the original data set includes historical power generation data, historical cloud layer images, and historical meteorological data of multiple meteorological parameters;

[0179] Generate a first training data set according to the original data set; wherein each first training data subset in the first training data set corresponds to an original data subset;

[0180] Using the first training data set to perform model training on the pre-built initial photovoltaic power prediction model to obtain a target photovoltaic power prediction model;

[0181] and / or,

[0182] Acquire target data; wherein the target data includes target cloud layer images corresponding to a plurality of consecutive target acquisition time points and target meteorological data of a plurality of meteorological parameters;

[0183] Based on the target photovoltaic power generation prediction model, the photovoltaic power generation prediction result corresponding to the target data is determined.

[0184] Figure 7 The internal structure diagram of the computer device of the embodiment of the present application is schematically shown. Figure 7 As shown, in one embodiment of the present application, a computer device is provided, which may be a terminal. The computer device includes a processor A01, a network interface A02, a display screen A04, an input device A05 and a memory (not shown in the figure) connected via a system bus. Among them, the processor A01 of the computer device is used to provide computing and control capabilities. The memory of the computer device includes an internal memory A03 and a non-volatile storage medium A06. The non-volatile storage medium A06 stores an operating system B01 and a computer program B02. The internal memory A03 provides an environment for the operation of the operating system B01 and the computer program B02 in the non-volatile storage medium A06. The network interface A02 of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor A01, a photovoltaic power generation power prediction model construction method and / or a photovoltaic power generation power prediction method are implemented. The display screen A04 of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device A05 of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device casing, or an external keyboard, touchpad or mouse.

[0185] Those skilled in the art will understand that Figure 7 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0186] In one embodiment, the photovoltaic power generation prediction model construction device / photovoltaic power generation prediction device provided by the present application can be implemented in the form of a computer program. The computer program can be Figure 7 The computer device shown in the figure is run. The memory of the computer device can store various program modules constituting the photovoltaic power generation prediction model building device / photovoltaic power generation prediction device, for example, Figure 4 The original data set acquisition module, training data set generation module and first prediction model generation module in the photovoltaic power generation prediction model construction device shown, and the computer program composed of each program module enable the processor to execute the steps in the photovoltaic power generation prediction model construction method of each embodiment of the present application described in this specification.

[0187] Figure 7 The computer device shown can be Figure 4The original data set acquisition module in the photovoltaic power generation power prediction model construction device shown in the figure performs the step A200, the training data set generation module performs the step A400 and the first prediction model generation module performs the step A600. Figure 5 The target data acquisition module in the photovoltaic power generation prediction device shown executes the step B200 and the photovoltaic power generation prediction module executes the step B400.

[0188] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0189] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes 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 generate 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.

[0190] 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 comprising 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.

[0191] 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.

[0192] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0193] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0194] Computer readable media include permanent and non-permanent, removable and non-removable media, and can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0195] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0196] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for constructing a photovoltaic power generation prediction model, characterized in that: The method includes: Acquire an original data set; wherein each original data subset in the original data set includes historical power generation data, historical cloud layer images, and historical meteorological data of multiple types of meteorological parameters; Generate a first training data set according to the original data set; wherein each first training data subset in the first training data set corresponds to an original data subset; The first training data set is used to perform model training on the pre-built initial photovoltaic power generation prediction model to obtain a target photovoltaic power generation prediction model.

2. The method according to claim 1, characterized in that Generate a first training data set according to the original data set, including: For each historical cloud layer image, annotate the historical cloud layer image to obtain an annotation result of the historical cloud layer image; wherein the annotation result of the historical cloud layer image includes historical cloud amount data corresponding to the historical cloud layer image; Generate a cloud amount data set based on each historical cloud layer image; wherein the historical cloud amount data corresponding to each historical cloud layer image is the first cloud amount data in the cloud amount data set; Based on the cloud cover dataset, determine the mean and variance corresponding to the cloud layer image; Determine normalized data corresponding to each first cloud amount data in the cloud amount data set according to the mean and variance corresponding to the cloud layer image; The first training data set includes normalized data corresponding to each first cloud amount data in the cloud amount data set.

3. The method according to claim 2, characterized in that The method further includes: The pre-built initial image quantization model is trained using the annotation results of each historical cloud image to obtain the target image quantization model.

4. The method according to claim 1, characterized in that Generate a first training data set according to the original data set, including: For each type of meteorological parameter, based on each historical meteorological data of the meteorological parameter, a first meteorological data set corresponding to the meteorological parameter is generated; wherein each historical meteorological data of the meteorological parameter is the first meteorological data in the first meteorological data set corresponding to the meteorological parameter; For each type of meteorological parameter, determine the mean and variance corresponding to the type of meteorological parameter according to the first meteorological data set corresponding to the type of meteorological parameter; For each type of meteorological parameter, determine the normalized data corresponding to each of the first meteorological data in the first meteorological data set corresponding to the meteorological parameter according to the mean and variance corresponding to the meteorological parameter; For each type of meteorological parameter, the first training data set includes normalized data corresponding to each first meteorological data in the first meteorological data set corresponding to the meteorological parameter of this type.

5. The method according to any one of claims 1 to 4, characterized in that The method further includes: The second original data is obtained according to the first preset time interval, and the following steps are performed: new cloud amount data corresponding to the new cloud layer image in the second original data is determined, and according to the new cloud amount data, whether the current target photovoltaic power generation prediction model meets the preset incremental training conditions is determined; if so, a second training data set is generated according to the original data set and each second original data, and the current target photovoltaic power generation prediction model is incrementally trained using the second training data set to obtain a new target photovoltaic power generation prediction model.

6. A photovoltaic power generation power prediction method, characterized in that: The method includes: Acquire target data; wherein the target data includes target cloud images corresponding to a plurality of consecutive target acquisition time points and target meteorological data of multiple types of meteorological parameters; Based on the target photovoltaic power prediction model according to any one of claims 1 to 5, a photovoltaic power prediction result corresponding to the target data is determined.

7. A photovoltaic power generation power prediction model construction device, characterized in that: The device includes: The original data set acquisition module is used to acquire the original data set; wherein each original data subset in the original data set includes historical power generation data, historical cloud layer images and historical meteorological data of multiple types of meteorological parameters; A training data set generation module, used to generate a first training data set according to the original data set; wherein each first training data subset in the first training data set corresponds to an original data subset; The first prediction model generation module is used to perform model training on a pre-built initial photovoltaic power generation prediction model using a first training data set to obtain a target photovoltaic power generation prediction model.

8. A photovoltaic power generation power prediction device, characterized in that: The device includes: A target data acquisition module is used to acquire target data; wherein the target data includes target cloud images corresponding to a plurality of consecutive target acquisition time points and target meteorological data of multiple types of meteorological parameters; A photovoltaic power generation prediction module is used to determine a photovoltaic power generation prediction result corresponding to target data based on the target photovoltaic power generation prediction model according to any one of claims 1 to 5.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the photovoltaic power generation prediction model construction method described in any one of claims 1 to 5 and / or the photovoltaic power generation prediction method described in claim 6 are implemented.

10. A machine-readable storage medium having instructions stored thereon, characterized in that: When the instruction is executed by a processor, the processor is configured to execute the photovoltaic power generation prediction model construction method described in any one of claims 1 to 5, and / or the photovoltaic power generation prediction method described in claim 6.

Images