Distributed photovoltaic output prediction method and device, equipment and medium

Abstract

The invention discloses a distributed photovoltaic output prediction method and device, equipment and a medium, and relates to the technical field of distributed photovoltaic, and the method comprises the steps: obtaining distributed photovoltaic output historical data corresponding to a prediction day, and outputting a basic prediction result of the distributed photovoltaic output of the prediction day through a prediction model; on the basis of the daily growth coefficient of the transformer area dust, predicting to obtain a predicted daily dust amount; and according to the basic prediction result of the distributed photovoltaic output of the prediction day and the dust amount of the prediction day, calculating to obtain a target prediction result of the distributed photovoltaic output of the prediction day. The method is suitable for a prediction scene of power distributed photovoltaic output.

Description

Technical Field

[0001] The present application relates to the field of distributed photovoltaic technology, and in particular to methods, devices, equipment, and media for predicting distributed photovoltaic output. Background Art

[0002] Distributed photovoltaic power generation is a new and promising method for power generation and comprehensive energy utilization. It promotes the principles of local generation, local grid connection, local conversion, and local use. Distributed photovoltaic power generation systems are typically installed in small spaces such as rooftops, walls, roads, and parking lots, or integrated with buildings. They are relatively small, with individual capacities typically ranging from a few kilowatts to tens of kilowatts. This power generation method not only reduces the load on the power grid, especially during peak photovoltaic power generation periods, but also enables immediate power generation and consumption, reducing losses associated with long-distance transmission and improving energy efficiency.

[0003] Existing distributed photovoltaic output prediction models do not take into account the personalized degree of dust volume in each substation or even the distributed photovoltaics under each substation, resulting in low accuracy of the final photovoltaic output prediction results. This may have an adverse impact on key links such as grid scheduling, energy storage system configuration and demand-side response, thereby affecting the overall performance and economic benefits of the distributed photovoltaic system. Summary of the Invention

[0004] In view of this, the present application provides a method, device, equipment and medium for predicting distributed photovoltaic output. The main purpose is to solve the technical problem that the distributed photovoltaic output prediction model in the existing technology does not take into account the personalized degree of dust amount of each substation or even each distributed photovoltaic under each substation, resulting in low accuracy of the final output photovoltaic output prediction result.

[0005] According to one aspect of the present application, a method for predicting distributed photovoltaic output is provided, the method comprising:

[0006] Obtain the historical data of distributed photovoltaic output corresponding to the forecast day, and use the forecast model to output the basic forecast results of distributed photovoltaic output on the forecast day;

[0007] Based on the daily growth coefficient of dust in the substation, the predicted daily dust amount is predicted;

[0008] According to the basic forecast results of distributed photovoltaic output on the forecast day and the dust amount on the forecast day, the target forecast results of distributed photovoltaic output on the forecast day are calculated.

[0009] According to another aspect of the present application, a distributed photovoltaic output prediction device is provided, the device comprising:

[0010] The first prediction module is used to obtain the historical data of distributed photovoltaic output corresponding to the prediction day and output the basic prediction result of distributed photovoltaic output on the prediction day using the prediction model;

[0011] The second prediction module is used to predict the daily dust volume based on the daily growth coefficient of the dust in the station area;

[0012] The third prediction module is used to calculate the target prediction result of the distributed photovoltaic output on the prediction day according to the basic prediction result of the distributed photovoltaic output on the prediction day and the dust amount on the prediction day.

[0013] According to another aspect of the present application, a computer storage medium is provided, on which a computer program is stored. When the program is executed by a processor, the above-mentioned distributed photovoltaic output prediction method is implemented.

[0014] According to another aspect of the present application, a computer device is provided, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor implements the above-mentioned distributed photovoltaic output prediction method when executing the program.

[0015] By means of the above technical solution, the distributed photovoltaic output prediction method, device, equipment and medium provided by this application are compared with the technical solution in the prior art in which the distributed photovoltaic output prediction model does not consider the personalized degree of dust amount of each substation or even each distributed photovoltaic under each substation, and does not consider the impact of dust on the power generation efficiency of photovoltaic modules. This application obtains the historical data of distributed photovoltaic output corresponding to the prediction day, and uses the prediction model to output the basic prediction result of the distributed photovoltaic output on the prediction day; based on the daily growth coefficient of the substation dust, the dust amount on the prediction day is predicted; based on the basic prediction result of the distributed photovoltaic output on the prediction day and the dust amount on the prediction day, the target prediction result of the distributed photovoltaic output on the prediction day is calculated. It can be seen that the distributed photovoltaic output prediction model with strong universality is used to obtain the basic prediction result of the distributed photovoltaic output. Further, based on the basic prediction of the distributed photovoltaic output, the dust shielding situation of the distributed photovoltaic is combined with the dust shielding of the distributed photovoltaic to obtain a more accurate prediction result of the distributed photovoltaic output, which provides a basis for grid scheduling and provides data support for energy storage systems and demand-side management, thereby optimizing the grid scheduling strategy and ensuring that photovoltaic power can be smoothly integrated into the large grid and achieved.

[0016] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0018] Figure 1 A schematic diagram of a process for predicting distributed photovoltaic output provided by an embodiment of the present application is shown;

[0019] Figure 2 A schematic diagram showing a flow chart of another method for predicting distributed photovoltaic output provided by an embodiment of the present application is shown;

[0020] Figure 3 A schematic diagram of the structure of a distributed photovoltaic output prediction device provided in an embodiment of the present application is shown;

[0021] Figure 4 A schematic structural diagram of another distributed photovoltaic output prediction device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0022] The present application will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that, unless there is a conflict, the embodiments and features in the embodiments of the present application can be combined with each other.

[0023] Aiming at the technical problem that the distributed photovoltaic output prediction model in the prior art does not take into account the individual degree of dust volume of each substation or even each distributed photovoltaic under each substation, resulting in low accuracy of the photovoltaic output prediction result outputted. This embodiment provides a distributed photovoltaic output prediction method, which obtains the basic target prediction result of distributed photovoltaic output based on a highly universal distributed photovoltaic output prediction model, and further combines the dust shielding situation of distributed photovoltaics on the basis of the basic prediction of distributed photovoltaic output to predict a more accurate distributed photovoltaic output prediction result. Figure 1 As shown, the above method includes the following steps:

[0024] Step 101: Obtain historical data of distributed photovoltaic output corresponding to a forecast day, and use a forecast model to output a basic forecast result of distributed photovoltaic output on the forecast day.

[0025] In this embodiment, based on the date characteristics of the forecast day, the historical date corresponding to the forecast day is obtained, thereby obtaining the corresponding historical distributed photovoltaic output data. This historical distributed photovoltaic output data is used as input data for the forecast model to obtain a basic forecast result for daily distributed photovoltaic output. This basic forecast result can, to a certain extent, accurately predict the photovoltaic power generation for a specific forecast day by utilizing high-precision historical meteorological data, historical power data, and advanced forecast models. In addition to the historical distributed photovoltaic output data, dust obstruction can also significantly reduce the power generation efficiency of photovoltaic modules to a certain extent. Therefore, this embodiment further calibrates the basic forecast result by considering the impact of dust obstruction on photovoltaic output, resulting in a more accurate forecast result.

[0026] Step 102: Based on the daily growth coefficient of the dust in the station area, the predicted daily dust amount is predicted.

[0027] In this embodiment, the dust volume on the predicted day is predicted without using the prediction model for the dust in the substation. This is different from the existing method of directly introducing dust influencing factors into the prediction model for overall training of the prediction model. Because the dust volume also includes the influence of uncertain factors such as rainfall and cleanliness, directly introducing the prediction model for training cannot achieve the effect of improving the accuracy of the prediction in practical applications, and it will cause unnecessary interference to the accuracy of the prediction results, which will in turn affect the accuracy of the prediction results. Therefore, this embodiment is based on the daily growth coefficient of the dust in the substation, and combined with the historical cleanliness data and historical rainfall data of the substation, to predict the dust volume on the predicted day, and then based on the basic prediction results of the distributed photovoltaic output on the predicted day and the dust volume on the predicted day, a more accurate prediction result is obtained.

[0028] Step 103: Calculate the target prediction result of the distributed photovoltaic output on the forecast day based on the basic prediction result of the distributed photovoltaic output on the forecast day and the dust amount on the forecast day.

[0029] In this embodiment, after using the forecast model to output a baseline forecast for distributed photovoltaic output on the forecasted day, a target forecast under the influence of that dust level is derived based on the baseline forecast and its corresponding dust level. This effectively improves the accuracy of the photovoltaic output forecast by fully accounting for the individualized effects of dust obstruction on each substation or distributed photovoltaic system.

[0030] In this embodiment, the above scheme can be followed to obtain historical distributed photovoltaic output data corresponding to the forecast day, and a prediction model can be used to output a basic forecast result for the distributed photovoltaic output on the forecast day. Based on the daily growth coefficient of dust in the substation, the dust volume on the forecast day is predicted. Based on the basic forecast result for the distributed photovoltaic output on the forecast day and the dust volume on the forecast day, a target forecast result for the distributed photovoltaic output on the forecast day is calculated. Compared to existing distributed photovoltaic output prediction models that fail to consider the individual degree of dust volume response for each substation, or even for each distributed photovoltaic under each substation, and fail to consider the impact of dust on the power generation efficiency of photovoltaic modules, this embodiment utilizes a highly universal distributed photovoltaic output prediction model to obtain a basic forecast result for distributed photovoltaic output. Furthermore, based on the basic forecast of distributed photovoltaic output, the dust shielding conditions of distributed photovoltaics are combined to obtain a more accurate forecast result for distributed photovoltaic output, providing a basis for grid scheduling and data support for energy storage systems and demand-side management, thereby optimizing grid scheduling strategies and ensuring that photovoltaic power can be smoothly integrated into the larger grid and absorbed.

[0031] Furthermore, as a refinement and expansion of the specific implementation of the above embodiment, in order to fully illustrate the specific implementation process of this embodiment, another distributed photovoltaic output prediction method is provided. This embodiment can solve the technical problem in the prior art that the distributed photovoltaic output prediction model does not take into account the personalized degree of dust amount of each substation or even each distributed photovoltaic under each substation, resulting in low accuracy of the final output photovoltaic output prediction result. This embodiment can effectively improve the output accuracy of the photovoltaic output prediction result by fully considering the photovoltaic output loss under the influence of personalized dust shielding of each substation or each distributed photovoltaic. Figure 2 As shown in the figure, taking the prediction scenario of distributed photovoltaic output as an example, the prediction methods of distributed photovoltaic output include:

[0032] Step 201: Train the prediction model.

[0033] To illustrate the specific implementation of step 201, as a preferred embodiment, step 201 includes: using distributed photovoltaic output historical data as model input data, training a machine learning basic model or a deep learning basic model, and obtaining a prediction model.

[0034] During implementation, the distributed photovoltaic output historical data includes historical meteorological data, historical power data and equipment data. The historical meteorological data includes at least light radiation intensity and temperature. The power data includes at least photovoltaic module characteristics (module quality, attenuation characteristics), module installation inclination and orientation, inverter conversion efficiency, and line loss. The equipment data includes at least substation capacity and load, and photovoltaic access location.

[0035] During implementation, the step of using the distributed photovoltaic output historical data as the model input feature includes: preprocessing the distributed photovoltaic output historical data to obtain model input data, specifically including: collecting historical meteorological data, historical power data and equipment data; performing linear interpolation and completion processing on the missing values ​​in the historical meteorological data and historical power data to obtain complete historical data; normalizing the complete historical data to obtain historical data quantization values, and the historical data quantization values ​​include historical meteorological data quantization values ​​and historical power data quantization values; performing trend feature extraction on the historical power data quantization values ​​to obtain trend items of historical power data; and using the historical meteorological data quantization values, historical power data trend items and equipment data as model input data.

[0036] During implementation, historical data on distributed photovoltaic output is obtained, including at least historical meteorological data, historical power data and equipment data that affect distributed photovoltaic output, and a training sample set is constructed based on the above historical data. The basic short-term prediction model is trained to obtain a short-term prediction model for predicting the distributed photovoltaic output results on a specific date. The above historical data is preprocessed to obtain a training sample set. The steps include: collecting and preprocessing distributed photovoltaic output historical data to obtain model input data. The preprocessing steps include: 1) Vacancy value processing: linear interpolation is performed on the missing values ​​in the historical meteorological data and historical power data. 2) Normalization processing: min-max normalization processing is performed on the completed historical meteorological data and historical power data, and the data is mapped to the range of 0-1. 3) Trend item extraction: the trend item of the completed historical power data is extracted. The trend item can be a time-based polynomial, a trigonometric polynomial, a seasonal mean, or a linear combination of the three.

[0037] The training steps of this basic short-term forecasting model include: the training sample set includes equipment data, historical meteorological data that has been completed and normalized, and historical power data that has been completed, normalized, and trend items extracted, and the preprocessed training sample set is used as the model input data; a machine learning model or a deep learning model, such as an LSTM neural network or an ARIMA-LSTM-DBN model, is selected; the model is trained based on the training sample set, model parameters are adjusted, and model performance is optimized.

[0038] Step 202: Using a clustering algorithm, determine the historical date based on the date attributes of the predicted day.

[0039] Step 203: Obtain the distributed photovoltaic output historical data of the historical date and use it as the distributed photovoltaic output historical data corresponding to the prediction day.

[0040] To illustrate the specific implementation of step 202, as a preferred embodiment, step 202 includes: obtaining an initial historical date corresponding to the date characteristics of the forecast day; obtaining a cluster space containing multiple cluster centers using Ward variance minimization based on the distributed photovoltaic output historical data of the initial historical date; projecting the date characteristics of the forecast day into the cluster space, determining the cluster center corresponding to the date characteristics of the forecast day, and using the date corresponding to the cluster center as the historical date corresponding to the date characteristics of the forecast day.

[0041] During implementation, the date characteristics of the forecast day, such as season, day type, and meteorological conditions corresponding to the season or day type, are obtained, and date characteristics similar to the forecast day are extracted from historical data to determine the initial historical date, and the distributed photovoltaic output historical data of the initial historical date is obtained; based on the improved K-means clustering algorithm and combined with the idea of ​​hierarchical clustering (Ward variance minimization), a clustering space containing multiple clustering centers is determined, and the multiple clustering centers are used to divide the obtained distributed photovoltaic output historical data of the initial historical date into several categories, so as to more accurately obtain a historical date that is closer to the date characteristics of the forecast day; the date characteristics of the forecast day are projected into the clustering space, the cluster center corresponding to the date characteristics of the forecast day is determined, and the historical date corresponding to the cluster center is used as the historical similar day, and the distributed photovoltaic output historical data of the historical similar day is used as the distributed photovoltaic output historical data corresponding to the forecast day.

[0042] Step 204: Use the prediction model to output a basic prediction result of the distributed photovoltaic output on the predicted day.

[0043] During implementation, historical distributed PV output data from similar days was used as model input. Based on the trained short-term forecast model, the model output was output as a forecast for similar days. This forecast served as the baseline for distributed PV output. Distributed PV output was represented by a daily power curve, representing the power generated at different points in the day. The training data used to train the short-term forecast model indicated that dust obstruction was not considered during model training. Therefore, the output forecast for distributed PV output represents the ideal output under a dust-free environment.

[0044] During implementation, the dust occlusion situation is considered separately and separated from the short-term prediction model, thereby further improving the predicted output result of the final output of this embodiment. Existing models do not consider dust occlusion, or reflect dust occlusion data in the training data, resulting in low accuracy of the prediction results. In this embodiment, the dust occlusion data is not directly added to the training sample set for training the basic short-term prediction model. On the one hand, this is to consider the simplification of model training data and clustering algorithm data, and to improve the rate of model training and algorithm clustering. On the other hand, this is because not all similar days in the substation will have the same dust occlusion situation, and the dust occlusion situation in the substation is usually displayed as a periodic change within the substation. For example, dust occlusion will change periodically with the usual cleaning day of the substation, and the dust occlusion intensity will also be affected by the environmental factors of the substation. It can be seen that the photovoltaic cleaning day is the personalized data of each distributed photovoltaic in the substation, and the dust occlusion degree is the personalized data within the substation, as well as the personalized data of each distributed photovoltaic under the substation, which is not universal. Therefore, dust occlusion data should not be trained in a universal basic short-term prediction model, nor should data clustering be performed in a clustering algorithm. Instead, adaptive changes should be made based on the predicted basic prediction results for different stations to improve the speed and the accuracy of the final output results.

[0045] Step 205: Based on the daily growth coefficient of the dust in the station area, the predicted daily dust amount is predicted.

[0046] In order to illustrate the specific implementation of step 205, as a preferred embodiment, a dust sensor is installed on the photovoltaic module of the distributed photovoltaic system, and the dust sensor is used to obtain the daily dust amount of the photovoltaic module. The step of calculating the daily growth coefficient of the dust in the substation includes: according to the obtained daily dust amount, the daily growth coefficient of the dust in the substation is calculated based on the total number of sampling days.

[0047] In order to illustrate the specific implementation of step 205, as a preferred embodiment, step 205 includes: determining the dust duration of the area from the predicted day based on the historical cleaning day or historical rainy day of the area closest to the predicted day; calculating the dust amount on the predicted day based on the daily growth coefficient of the dust in the area and the dust duration of the area.

[0048] During implementation, the daily growth coefficient of the dust in the substation, the historical cleaning data of the substation, and the historical rainfall data of the substation are obtained; among them, the historical cleaning data of the substation is the historical cleaning record, which is used to obtain the historical cleaning day, and the historical rainfall data is the historical rainfall day and its corresponding rainfall. Based on the daily growth coefficient of the dust in the substation, the dust amount Xβ on the predicted day is obtained from the most recent historical cleaning day or historical rainfall day of the substation before the predicted day, where X is the interval between the most recent historical cleaning day or historical rainfall day of the substation and the predicted day. To ensure the accurate prediction of the dust amount on the predicted day, the dates with less precipitation such as light rain or plum rain, which have negligible effects on dust cleaning, are ignored, that is, the rainfall day is the date when the rainfall reaches the set threshold. The steps for obtaining the daily growth coefficient of the dust in the substation include:

[0049] A dust sensor is installed on the side or top of any distributed photovoltaic module in the area to obtain dust obstruction data of the photovoltaic module on a daily basis. The daily growth coefficient is calculated as follows:

[0050]

[0051] Where β is the daily growth coefficient, D i+1 is the amount of dust on day i+1, D i is the dust amount on the i-th day, and n is the total number of sampling days.

[0052] Step 206: Based on the predicted daily dust amount, the power generation loss percentage corresponding to the predicted daily dust amount is determined using a relationship function between the dust amount and the power generation loss percentage.

[0053] To illustrate the specific implementation of step 206, as a preferred embodiment, before step 206, the method further includes: obtaining the dust amount of the photovoltaic module and the output power of the photovoltaic module on a daily basis, and establishing a relationship function between the dust amount and the power generation loss percentage using a regression algorithm.

[0054] Step 207: Calculate a target forecast result of the distributed photovoltaic output on the forecast day based on the basic forecast result of the distributed photovoltaic output on the forecast day and the power generation loss percentage on the forecast day.

[0055] During implementation, based on the operation and maintenance records of the photovoltaic system, or by installing dust sensors and power monitoring equipment to collect the output power data of the photovoltaic modules and the corresponding dust amount data in real time, regression analysis methods such as linear regression and polynomial regression are used to establish a relationship function between the dust amount and the power generation loss percentage. According to the relationship function between the dust amount and the power generation loss percentage on the predicted day, the power generation loss percentage corresponding to the current dust amount is calculated, and then the power generation loss percentage and the basic prediction result are used as the target prediction result of the distributed photovoltaic output on the predicted day.

[0056] By applying the technical solution of this embodiment, historical meteorological data, historical power data, and equipment data that affect distributed photovoltaic output are collected, and a basic short-term prediction model is constructed and trained based on these data; based on the clustering algorithm, historical similar days of the prediction day are determined, and the data of the historical similar days are used as model input data, and the basic prediction result of the distributed photovoltaic output on the prediction day is output based on the trained basic short-term prediction model; the daily growth coefficient of the dust in the substation, the historical clean data of the substation, and the historical rainfall data of the substation are obtained; based on the daily growth coefficient of the dust in the substation, the dust amount on the prediction day is obtained from the most recent historical clean day or historical rainfall day in the substation before the prediction day; based on the basic prediction result of the distributed photovoltaic output on the prediction day and the dust amount on the prediction day, the target prediction result under the influence of the dust amount on the prediction day is obtained. Compared with the technical solution of the existing distributed photovoltaic output prediction scheme that does not take into account the personalized dust shielding of the distributed photovoltaic in the substation, this embodiment can fully consider the photovoltaic output loss under the influence of the personalized dust shielding of each substation or each distributed photovoltaic, thereby improving the accuracy of the photovoltaic output prediction result.

[0057] Further, as Figure 1 The specific implementation of the method, the embodiment of the present application provides a distributed photovoltaic output prediction device, such as Figure 3 As shown, the device includes: a first prediction module 32, a second prediction module 33, and a third prediction module 34.

[0058] The first prediction module 32 is used to obtain the historical data of distributed photovoltaic output corresponding to the prediction day, and output the basic prediction result of the distributed photovoltaic output on the prediction day using the prediction model.

[0059] The second prediction module 33 is used to predict the daily dust volume based on the daily growth coefficient of the dust in the station area.

[0060] The third prediction module 34 is configured to calculate a target prediction result of the distributed photovoltaic output on the predicted day based on the basic prediction result of the distributed photovoltaic output on the predicted day and the dust amount on the predicted day.

[0061] In a specific application scenario, the distributed photovoltaic output historical data includes historical meteorological data, historical power data and equipment data. The historical meteorological data includes at least light radiation intensity and temperature. The power data includes at least photovoltaic module characteristics, module installation inclination and orientation, inverter conversion efficiency, and line loss. The equipment data includes at least substation capacity and load, and photovoltaic access location.

[0062] In specific application scenarios, such as Figure 4 As shown, the device further includes a training module 31 , and the first prediction module 32 includes: an acquisition submodule 321 .

[0063] The acquisition submodule 321 is configured to determine a historical date based on the date characteristics of the prediction date using a clustering algorithm; and to acquire the distributed photovoltaic output historical data of the historical date as the distributed photovoltaic output historical data corresponding to the prediction date.

[0064] In a specific application scenario, the acquisition submodule 321 is specifically used to: obtain the initial historical date corresponding to the date characteristics of the prediction day; and, based on the distributed photovoltaic output historical data of the initial historical date, use Ward variance minimization to obtain a cluster space containing multiple cluster centers; and, project the date characteristics of the prediction day into the cluster space, determine the cluster center corresponding to the date characteristics of the prediction day, and use the date corresponding to the cluster center as the historical date corresponding to the prediction day.

[0065] In a specific application scenario, a dust sensor is installed on a photovoltaic module of a distributed photovoltaic system. The dust sensor is used to obtain the daily dust amount of the photovoltaic module. The second prediction module 33 includes: a first calculation submodule 331 and a second calculation submodule 332 .

[0066] The calculation submodule 331 is used to calculate the daily growth coefficient of the dust in the area based on the acquired daily dust amount and the total number of sampling days.

[0067] The second calculation submodule 332 is used to determine the dust duration of the area from the predicted day based on the historical cleaning day or historical rainy day of the area closest to the predicted day; and to calculate the dust amount on the predicted day based on the daily growth coefficient of the dust in the area and the dust duration of the area.

[0068] In a specific application scenario, the third prediction module 34 is further used to: obtain the dust amount of the photovoltaic module and the output power of the photovoltaic module on a daily basis, and establish a relationship function between the dust amount and the power generation loss percentage using a regression algorithm.

[0069] In a specific application scenario, the third prediction module 34 includes: a first determination submodule 341 and a second determination submodule 342 .

[0070] The first determining submodule 341 is configured to determine the power generation loss percentage corresponding to the predicted daily dust amount by using a relationship function between the dust amount and the power generation loss percentage according to the predicted daily dust amount.

[0071] The second determination submodule 342 is configured to calculate a target prediction result of the distributed photovoltaic output on the prediction day according to the basic prediction result of the distributed photovoltaic output on the prediction day and the power generation loss percentage on the prediction day.

[0072] In a specific application scenario, the training module 31 is used to train a prediction model.

[0073] The training module 31 is specifically used to use the distributed photovoltaic output historical data as model input data to train the machine learning basic model or the deep learning basic model to obtain a prediction model.

[0074] In a specific application scenario, the training module 31 includes: a preprocessing submodule 311 .

[0075] The preprocessing submodule 311 is used to preprocess the historical data of distributed photovoltaic output to obtain model input data, specifically for collecting historical meteorological data, historical power data and equipment data; performing linear interpolation to complete the missing values ​​in the historical meteorological data and historical power data to obtain complete historical data; normalizing the complete historical data to obtain historical data quantization values, which include historical meteorological data quantization values ​​and historical power data quantization values; performing trend feature extraction on the historical power data quantization values ​​to obtain trend items of the historical power data; and using the historical meteorological data quantization values, historical power data trend items and equipment data as model input data.

[0076] It should be noted that for other corresponding descriptions of the functional units involved in the distributed photovoltaic output prediction device provided in the embodiment of the present application, please refer to Figure 1 and Figure 2 The corresponding description in will not be repeated here.

[0077] Based on the above Figure 1 and Figure 2 The method shown in FIG. 1 is a method for performing the above-mentioned operation. Accordingly, the embodiment of the present application further provides a computer storage medium having a computer program stored thereon, which implements the above-mentioned operation when the computer program is executed by the processor. Figure 1 and Figure 2 The distributed photovoltaic output prediction method shown in FIG.

[0078] Based on this understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a storage medium (which can be a CD-ROM, USB flash drive, mobile hard disk, etc.), including a number of instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods described in each implementation scenario of the present application.

[0079] Based on the above Figure 1 、 Figure 2 The method shown, and Figure 3 、 Figure 4 In order to achieve the above-mentioned purpose, the embodiment of the present application further provides a computer device, which can be a personal computer, a server, a network device, etc. The physical device includes a storage medium and a processor; the storage medium is used to store a computer program; the processor is used to execute the computer program to achieve the above-mentioned Figure 1 and Figure 2 The distributed photovoltaic output prediction method shown in FIG.

[0080] Optionally, the computer device may further include a user interface, a network interface, a camera, a radio frequency (RF) circuit, a sensor, an audio circuit, a Wi-Fi module, etc. The user interface may include a display, an input unit such as a keyboard, etc., and may optionally include a USB interface, a card reader interface, etc. The network interface may optionally include a standard wired interface, a wireless interface (such as a Bluetooth interface, a Wi-Fi interface), etc.

[0081] Those skilled in the art will understand that the computer device structure provided in this embodiment does not constitute a limitation on the physical device, and may include more or fewer components, or a combination of certain components, or different component arrangements.

[0082] The storage medium may also include an operating system and a network communication module. An operating system is a program that manages the hardware and software resources of a computer device, supporting the execution of information processing programs and other software and / or programs. The network communication module facilitates communication between components within the storage medium, as well as with other hardware and software within the physical device.

[0083] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present application can be implemented by means of software plus the necessary general hardware platform, or by hardware. By applying the technical solution of the present application, compared with the technical solution in the prior art in which the distributed photovoltaic output prediction model does not take into account the individual degree of dust volume of each substation or even each distributed photovoltaic under each substation, and does not take into account the impact of dust on the power generation efficiency of photovoltaic modules, the present embodiment can use a highly universal distributed photovoltaic output prediction model to obtain the basic prediction results of distributed photovoltaic output, and further combine the dust shielding of distributed photovoltaics on the basis of the basic prediction of distributed photovoltaic output to obtain a more accurate prediction result of distributed photovoltaic output, provide a basis for grid scheduling, and provide data support for energy storage systems and demand-side management, thereby optimizing grid scheduling strategies and ensuring that photovoltaic power can be smoothly integrated into the large grid and absorbed.

[0084] Those skilled in the art will understand that the accompanying drawings are only schematic diagrams of a preferred implementation scenario, and the modules or processes in the accompanying drawings are not necessarily required to implement the present application. Those skilled in the art will understand that the modules in the devices in the implementation scenario can be distributed in the devices of the implementation scenario according to the implementation scenario description, or can be changed accordingly and located in one or more devices different from the implementation scenario. The modules of the above-mentioned implementation scenario can be combined into one module, or can be further split into multiple sub-modules.

[0085] The serial numbers of the above application are for descriptive purposes only and do not represent the advantages or disadvantages of the implementation scenarios. The above disclosure only discloses several specific implementation scenarios of the present application, but the present application is not limited thereto. Any changes that can be conceived by those skilled in the art should fall within the scope of protection of the present application.

Claims

1. A method for predicting distributed photovoltaic output, characterized in that: include: Obtain the historical data of distributed photovoltaic output corresponding to the forecast day, and use the forecast model to output the basic forecast results of distributed photovoltaic output on the forecast day; Based on the daily growth coefficient of dust in the substation, the predicted daily dust amount is predicted; According to the basic forecast results of distributed photovoltaic output on the forecast day and the dust amount on the forecast day, the target forecast results of distributed photovoltaic output on the forecast day are calculated.

2. The method according to claim 1, characterized in that The distributed photovoltaic output historical data includes historical meteorological data, historical power data and equipment data. The historical meteorological data includes at least light radiation intensity and temperature. The power data includes at least photovoltaic module characteristics, module installation inclination and orientation, inverter conversion efficiency, and line loss. The equipment data includes at least substation capacity and load, and photovoltaic access location.

3. The method according to claim 1, characterized in that The step of obtaining the distributed photovoltaic output historical data corresponding to the forecast day includes: Using clustering algorithms, historical dates are determined based on the date characteristics of the predicted date; The distributed photovoltaic output historical data of the historical date is obtained and used as the distributed photovoltaic output historical data corresponding to the prediction day.

4. The method according to claim 3, characterized in that The step of using a clustering algorithm to determine the historical date according to the date characteristics of the predicted day includes: Obtaining an initial historical date corresponding to the date feature of the predicted day; Based on the historical data of distributed photovoltaic output on the initial historical date, a cluster space containing multiple cluster centers is obtained by minimizing the Ward variance. The date features of the predicted day are projected into the cluster space, the cluster center corresponding to the date features of the predicted day is determined, and the date corresponding to the cluster center is used as the historical date corresponding to the predicted day.

5. The method according to claim 1, wherein A dust sensor is installed on the photovoltaic module of the distributed photovoltaic system. The dust sensor is used to obtain the daily dust amount of the photovoltaic module. The steps of calculating the daily growth coefficient of the dust in the station area include: According to the daily dust amount obtained, the daily growth coefficient of dust in the substation area is calculated based on the total number of sampling days; The step of predicting the daily dust amount based on the daily growth coefficient of the dust in the station area includes: Determine the dust duration of the area from the forecast date based on the historical clean days or historical rainy days closest to the forecast date; Based on the daily growth coefficient of dust in the substation area and the duration of dust in the substation area, the predicted daily dust volume is calculated.

6. The method according to claim 1 or 5, characterized in that Before the step of calculating the target prediction result of the distributed photovoltaic output on the predicted day based on the basic prediction result of the distributed photovoltaic output on the predicted day and the dust amount on the predicted day, the method further includes: By obtaining the dust content of photovoltaic modules and the output power of photovoltaic modules on a daily basis, a regression algorithm is used to establish a relationship function between dust content and power generation loss percentage. The step of calculating a target prediction result of distributed photovoltaic output on the predicted day based on the basic prediction result of distributed photovoltaic output on the predicted day and the predicted dust amount on the predicted day includes: According to the predicted daily dust amount, the power generation loss percentage corresponding to the predicted daily dust amount is determined using the relationship function between the dust amount and the power generation loss percentage; The target forecast result of the distributed photovoltaic output on the forecast day is calculated based on the basic forecast result of the distributed photovoltaic output on the forecast day and the power generation loss percentage on the forecast day.

7. The method according to claim 1 or 2, characterized in that The training step of the prediction model includes: Use the historical data of distributed photovoltaic output as model input data to train the machine learning basic model or deep learning basic model to obtain a prediction model; The step of using the distributed photovoltaic output historical data as the model input feature includes: preprocessing the distributed photovoltaic output historical data to obtain model input data, specifically including: Collect historical meteorological data, historical power data and equipment data; Perform linear interpolation to fill in the missing values ​​in historical meteorological data and historical power data to obtain complete historical data; Normalizing the complete historical data to obtain historical data quantization values, wherein the historical data quantization values ​​include historical meteorological data quantization values ​​and historical power data quantization values; Extract trend features from the quantified values ​​of historical power data to obtain trend items of the historical power data; The quantitative values ​​of historical meteorological data, trend items of historical power data and equipment data are used as model input data.

8. A distributed photovoltaic output prediction device, characterized in that: include: The first prediction module is used to obtain the historical data of distributed photovoltaic output corresponding to the prediction day and output the basic prediction result of distributed photovoltaic output on the prediction day using the prediction model; The second prediction module is used to predict the daily dust volume based on the daily growth coefficient of the dust in the station area; The third prediction module is used to calculate the target prediction result of the distributed photovoltaic output on the prediction day according to the basic prediction result of the distributed photovoltaic output on the prediction day and the dust amount on the prediction day.

9. A computer storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for predicting distributed photovoltaic output according to any one of claims 1 to 7 is implemented.

10. A computer device comprising a storage medium, a processor, and a computer program stored in the storage medium and executable on the processor, wherein: When the processor executes the program, the distributed photovoltaic output prediction method according to any one of claims 1 to 7 is implemented.

Images