Photovoltaic power generation power short-term prediction method, device, equipment and medium

Abstract

The invention relates to the technical field of optical power prediction, and discloses a photovoltaic power generation power short-term prediction method and device, equipment and a medium, and the method comprises the steps: obtaining photovoltaic power data and irradiance data of a day before a target prediction day; inputting the photovoltaic power data into the first model and the second model; obtaining a first prediction sequence and a second prediction sequence of the photovoltaic power; inputting the irradiance data into a day-ahead weather state pre-judgment model to obtain a weather state pre-judgment result of the target prediction day; and based on the weather state pre-judgment result, selecting a target fusion mode, and fusing the first prediction sequence and the second prediction sequence to obtain a photovoltaic power generation power short-term prediction result. According to the technical scheme provided by one or more embodiments of the invention, the LSTM-RNN model and the time correlation model are introduced, and day-ahead weather state pre-judgment is combined, so that the effect of improving the accuracy of photovoltaic power prediction is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of optical power prediction, and in particular to a method, device, equipment and medium for short-term prediction of photovoltaic power generation. Background Art

[0002] Traditional artificial neural networks (ANN), recurrent neural networks (RNN) and long short-term memory networks (LSTM) have important applications in the field of time series prediction, but there are some defects. When dealing with time series problems, artificial neural networks (ANN) only establish nonlinear mappings between inputs and outputs, ignoring the temporal correlation between data, which is insufficient in scenarios such as photovoltaic power prediction that need to consider time changes. In addition, recurrent neural networks (RNN) are prone to gradient vanishing problems during training, resulting in reduced sensitivity to early input information and affecting the ability to learn long-term dependencies. At the same time, related technologies often do not fully consider the impact of weather conditions when predicting photovoltaic power generation, and changes in weather conditions have a significant impact on the output power of photovoltaic power stations.

[0003] In view of this, it is necessary to improve the accuracy of photovoltaic power prediction. Summary of the invention

[0004] In view of this, the present invention provides a method, device, equipment and medium for short-term prediction of photovoltaic power generation to solve the problem of insufficient accuracy of photovoltaic power prediction.

[0005] In a first aspect, the present invention provides a method for short-term prediction of photovoltaic power generation, the method comprising: obtaining photovoltaic power data and irradiance data of the day before a target prediction day; inputting the photovoltaic power data into a first model and a second model respectively; obtaining a first prediction sequence and a second prediction sequence of photovoltaic power; inputting the irradiance data into a day-ahead weather condition prediction model to obtain a weather condition prediction result of the target prediction day; based on the weather condition prediction result, selecting a target fusion method, fusing the first prediction sequence with the second prediction sequence, and obtaining a short-term prediction result of photovoltaic power generation.

[0006] In the second aspect, the present invention provides a short-term prediction device for photovoltaic power generation, which includes: a data acquisition module for acquiring photovoltaic power data and irradiance data of the day before a target prediction day; a sequence acquisition module for inputting the photovoltaic power data into a first model and a second model respectively; obtaining a first prediction sequence and a second prediction sequence of photovoltaic power; a weather prediction module for inputting the irradiance data into a weather state prediction model before the day to obtain a weather state prediction result of the target prediction day; a sequence fusion module for selecting a target fusion method based on the weather state prediction result, fusing the first prediction sequence with the second prediction sequence, and obtaining a short-term prediction result of photovoltaic power generation.

[0007] In a third aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor are communicatively connected to each other, computer instructions are stored in the memory, and the processor executes the short-term prediction method for photovoltaic power generation according to the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0008] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the photovoltaic power short-term prediction method of the first aspect or any corresponding embodiment thereof.

[0009] In a fifth aspect, the present invention provides a computer program product, comprising computer instructions for causing a computer to execute the photovoltaic power short-term prediction method of the first aspect or any corresponding embodiment thereof.

[0010] The technical solution provided by one or more embodiments of the present invention first obtains the photovoltaic power and irradiance data of the day before the target prediction day, and then inputs these data into two different prediction models: the first model and the second model. The two models output the first prediction sequence and the second prediction sequence of photovoltaic power respectively. Then, the irradiance data is used to input into the weather state prediction model of the day before to obtain the weather state prediction result of the target prediction day. According to this prediction result, the most appropriate fusion method is selected, and the first prediction sequence and the second prediction sequence are fused to finally obtain the short-term prediction result of photovoltaic power generation.

[0011] It can be seen that the technical solution provided by the present invention comprehensively considers historical photovoltaic power data and irradiance data, as well as the impact of weather changes on photovoltaic power generation, thereby improving the accuracy and reliability of the prediction and achieving more accurate short-term photovoltaic power prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0013] Figure 1 is a schematic flow chart of a method for short-term prediction of photovoltaic power generation according to an embodiment of the present invention;

[0014] Figure 2 is a structural block diagram of a photovoltaic power short-term prediction device according to an embodiment of the present invention;

[0015] Figure 3 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0017] The embodiment of the present invention provides a method for short-term prediction of photovoltaic power generation, which improves the accuracy of photovoltaic power prediction by introducing an LSTM-RNN model and a time association model and combining the day-ahead weather state prediction.

[0018] According to an embodiment of the present invention, an embodiment of a method for short-term prediction of photovoltaic power generation is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0019] In this embodiment, a method for short-term prediction of photovoltaic power generation is provided. Figure 1 is a flow chart of a method for short-term prediction of photovoltaic power generation according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0020] Step S101, obtaining photovoltaic power data and irradiance data of the day before the target prediction day.

[0021] Photovoltaic power data refers to the actual power data generated by the photovoltaic power station one day before the target forecast date. Irradiance data: Irradiance refers to the intensity of solar radiation reaching the surface of the earth. These data can be collected in real time through monitoring and recording equipment. These data are usually automatically collected by the monitoring system of the photovoltaic power station and stored in the database for analysis and forecasting. Obtaining photovoltaic power and irradiance data one day before the target forecast date is the basis for realizing short-term photovoltaic power forecasting, providing the model with necessary historical information and environmental condition data for more accurate forecasting.

[0022] Step S102, inputting photovoltaic power data into the first model and the second model respectively; obtaining a first prediction sequence and a second prediction sequence of photovoltaic power.

[0023] The first model and the second model analyze the input photovoltaic power data according to their algorithms and internal mechanisms, and generate a series of prediction values, which constitute the so-called "first prediction sequence" and "second prediction sequence". The prediction sequence output by each model is a series of estimated values ​​of future photovoltaic power generation. The obtained first prediction sequence and second prediction sequence will be fused in subsequent steps. The fusion method can be a simple weighted average or a more complex optimization algorithm, the purpose of which is to find an optimal combination so that the final prediction result is closest to the actual photovoltaic power generation.

[0024] Step S103, inputting the irradiance data into a weather condition prediction model for the day before, and obtaining a weather condition prediction result for the target forecast day.

[0025] The weather prediction model is used to analyze irradiance data to predict weather conditions for the next day or days. The model predicts the possible weather conditions on the target forecast day, such as sunny, cloudy, overcast or rainy.

[0026] Step S104, based on the weather state prediction result, a target fusion method is selected to fuse the first prediction sequence with the second prediction sequence to obtain a short-term prediction result of photovoltaic power generation.

[0027] Through the day-ahead weather state prediction model, we get the weather condition forecast for the target prediction day, such as sunny, cloudy or overcast. The fusion strategy can be simple, such as weighted average, or complex, such as multi-model integration based on optimization algorithm. The goal of fusion is to find an optimal weight combination so that the final prediction result is closest to the actual photovoltaic power generation power. According to the weather state prediction results, the weights of the two prediction sequences are intelligently adjusted. By fusing the two prediction sequences, a short-term prediction result of photovoltaic power generation power that comprehensively considers multiple factors is obtained. This result should be able to better reflect the actual power generation situation and improve the accuracy and reliability of the prediction.

[0028] The short-term photovoltaic power generation prediction method provided in this embodiment can significantly improve the accuracy and adaptability of the prediction by integrating historical data, multi-model prediction results and real-time weather status prediction, and can adapt to the uncertainty brought about by weather changes, thereby providing more accurate and reliable support for the power generation prediction of photovoltaic power stations.

[0029] In some optional implementations, the first model and the second model in the above step S102 include: the first model is a LSTM-RNN model;

[0030] The second model includes a reference matrix constructed at a specific prediction moment based on historical photovoltaic power data; the second model is used to predict the power generation on the target prediction day based on the data of the same date and adjacent dates in history.

[0031] The traditional RNN model is prone to gradient vanishing during the training process, that is, as the input sequence of the new time step enters the RNN, the sensitivity of the RNN to the input information of the old time step gradually decreases. The long short-term memory (LSTM) model can effectively solve the shortcomings of RNN and better store and retrieve information. Multiple units are introduced in the LSTM network, including input gate, output gate, forget gate and memory unit. Multiple gates control the state variables of the memory unit to store and retrieve information for a long time. When the input gate is activated, the input information will be stored in the memory unit. When the output gate is activated, the information in the memory unit will be passed to the next level of neurons. When the forget gate is activated, the information stored in the memory unit will be cleared. In this way, LSTM can perform long-term and short-term memory on the input sequence. Similar to RNN, LSTM is also trained through the forward process and the reverse process. The process of calculating the gradient backward to update the weight is called BPTT, which is composed of the chain rule. In short, the loss function not only receives the residual returned by the output layer of the current time, but also needs to accept the residual returned by the hidden layer of the next time step, and then update the weight.

[0032] Due to the influence of the earth's revolution around the sun, the extraterrestrial irradiance at a specific geographical location shows obvious time correlation characteristics, which are mainly reflected in annual periodicity and near similarity. Here, annual periodicity means that due to the same relative positions of the earth and the sun, the extraterrestrial irradiance at the same location in different years and the same accumulated days should be the same; near similarity means that due to the fact that the relative positions of the earth and the sun have not changed basically, the irradiance data of adjacent days should be close to each other. Under sunny conditions, the extraterrestrial irradiance is highly correlated with the surface irradiance, so the photovoltaic power generation power will also show similar annual periodicity and near similarity. The reference value of photovoltaic power generation power can be calculated based on the data of the same date and adjacent dates in history.

[0033] In some optional implementations, before the above step S102, the process includes:

[0034] Step a11, obtaining a periodicity scale parameter and a similarity scale parameter; the periodicity scale parameter is used to indicate the number of years of historical data selected forward from the prediction time point; the similarity scale parameter is used to indicate the number of days of historical data adjacent to the prediction time point.

[0035] Step a12, constructing a reference matrix based on the periodicity scale parameter, the similarity scale parameter and the historical photovoltaic power data.

[0036] Step a13, using the reference matrix to calculate a reference value, the reference value represents the photovoltaic power characteristics of the same date and adjacent dates in history.

[0037] To determine the data required for annual periodic correction, define the periodic scale parameter L, that is, the number of years of historical data selected forward from the prediction time point; to determine the data required for adjacent similarity correction, define the similarity scale parameter T, that is, the number of days of historical data selected forward / backward from the prediction time point. Both T and L are positive integers. According to the determined scale parameter values, the reference matrix R of photovoltaic power generation at the characteristic prediction moment can be calculated from the historical photovoltaic power data. Taking T = 3, L = 2 as an example, the reference matrix R is shown as follows:

[0038]

[0039] where r i,j It is the photovoltaic power at the jth day (j<0) before the i-th time before the prediction time or the j-th day (j>0) after the prediction time (i=1, 2, 3; j=-2, -1, 0, 1, 2).

[0040] Then, according to the correlation reference matrix R, the photovoltaic power reference value can be further calculated according to the following formula, which is only determined by the selected periodicity scale parameter and similarity scale parameter.

[0041] In some optional implementations, step S103 includes:

[0042] Step b11, based on the irradiance data, using the irradiance short-term prediction model, obtains the irradiance prediction result of the target prediction day.

[0043] Step b12, based on the irradiance prediction result, using the weather state classification recognition model, obtains the weather state prediction result of the target prediction day. In some optional implementations, before step S103, the process includes:

[0044] Step c11, receiving historical daily irradiance data.

[0045] Step c12, based on the historical daily irradiance data, determine the corresponding daily weather state label.

[0046] Step c13, training a weather state prediction model based on the weather state label and the corresponding irradiance data.

[0047] A clustering algorithm is used to divide different daily weather states according to the daily irradiance data of historical photovoltaic power stations; secondly, the weather state classification and recognition model is trained according to the clustering results, that is, the daily weather state label and the corresponding daily irradiance curve data; a photovoltaic power station power generation power day-ahead prediction model based on different algorithm principles is established to predict the power of the next day according to the power data of the previous day, and to predict the irradiance data of the next day according to the irradiance data of the previous day. Algorithms that can be used include ANN, SVM, ARIMA, etc.; the weather state of the forecast day is judged according to the prediction result data of each general model and the weather state classification and recognition model, and the final weather state of the forecast day is determined by voting.

[0048] In this embodiment, a photovoltaic power generation short-term prediction device is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware of a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0049] This embodiment provides a photovoltaic power short-term prediction device. Figure 2 As shown, including:

[0050] The data acquisition module 201 is used to acquire the photovoltaic power data and irradiance data of the day before the target prediction day;

[0051] The sequence acquisition module 202 is used to input the photovoltaic power data into the first model and the second model respectively to obtain the first prediction sequence and the second prediction sequence of the photovoltaic power;

[0052] The weather prediction module 203 is used to input the irradiance data into the weather state prediction model of the previous day to obtain the weather state prediction result of the target prediction day;

[0053] The sequence fusion module 204 is used to select a target fusion method based on the weather state prediction result, fuse the first prediction sequence with the second prediction sequence, and obtain a short-term prediction result of photovoltaic power generation.

[0054] In some optional embodiments, the device further comprises:

[0055] A model selection module, where the first model is a LSTM-RNN model;

[0056] The second model includes a reference matrix constructed at a specific prediction time based on historical photovoltaic power data; the second model is used to predict the power generation on the target prediction day based on the data of the same date and adjacent dates in history;

[0057] The reference value calculation module is also used to obtain a periodicity scale parameter and a similarity scale parameter; the periodicity scale parameter is used to indicate the number of years of historical data selected forward from the prediction time point; the similarity scale parameter is used to indicate the number of days of adjacent historical data selected from the prediction time point;

[0058] Based on the periodicity scale parameter, similarity scale parameter and historical PV power data, a reference matrix is ​​constructed;

[0059] The reference matrix is ​​used to calculate a reference value, wherein the reference value represents the photovoltaic power characteristics of the same date and adjacent dates in history. In some optional embodiments, the device further includes:

[0060] The irradiance prediction module is used to obtain the irradiance prediction result of the target prediction day based on the irradiance data and using the irradiance short-term prediction model;

[0061] Based on the irradiance prediction results, the weather state classification and recognition model is used to obtain the weather state prediction results for the target prediction day;

[0062] Model training module, used to receive historical daily irradiance data;

[0063] Based on the historical daily irradiance data, determine the corresponding daily weather state label;

[0064] Based on the weather state labels and the corresponding irradiance data, a weather state prediction model is trained.

[0065] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0066] The photovoltaic power generation short-term prediction device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0067] The embodiment of the present invention also provides a computer device having the above Figure 2 The short-term prediction device for photovoltaic power generation is shown.

[0068] See also Figure 3 , Figure 3 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 3As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 3 A processor 10 is taken as an example.

[0069] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0070] The memory 20 stores instructions executable by at least one processor 10, so that at least one processor 10 executes the method shown in the above embodiment.

[0071] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0072] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0073] The computer device also includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Figure 3 The example of connecting through bus is taken in the following.

[0074] The input device 30 can receive input digital or character information, and generate key signal input related to the user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator bar, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 may include a display device, an auxiliary lighting device (e.g., an LED) and a tactile feedback device (e.g., a vibration motor), etc. The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display and a plasma display. In some optional embodiments, the display device can be a touch screen.

[0075] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0076] A part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the existence of the computer program instruction in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc., and accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium accessible to the computer.

[0077] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for short-term prediction of photovoltaic power generation, characterized in that: The method comprises: Obtain the photovoltaic power data and irradiance data of the day before the target forecast day; Inputting the photovoltaic power data into the first model and the second model respectively; obtaining a first prediction sequence and a second prediction sequence of photovoltaic power; Inputting the irradiance data into a weather condition prediction model for the day before to obtain a weather condition prediction result for the target forecast day; Based on the weather state prediction result, a target fusion mode is selected, and the first prediction sequence is fused with the second prediction sequence to obtain a short-term prediction result of photovoltaic power generation.

2. The method according to claim 1, characterized in that The first model is an LSTM-RNN model; The second model includes a reference matrix constructed at a specific prediction moment based on historical photovoltaic power data; the second model is used to predict the power generation on the target prediction day based on historical data of the same date and adjacent dates.

3. The method according to claim 2, characterized in that Before inputting the photovoltaic power data into the first model and the second model respectively, the method further includes: Obtaining a periodicity scale parameter and a similarity scale parameter; the periodicity scale parameter is used to indicate the number of years of historical data selected forward from the prediction time point; the similarity scale parameter is used to indicate the number of days of historical data adjacent to the prediction time point; Based on the periodicity scale parameter, the similarity scale parameter and the historical photovoltaic power data, a reference matrix is ​​constructed; The reference value is calculated using the reference matrix, and the reference value represents the photovoltaic power characteristics of the same date and adjacent dates in history.

4. The method according to any one of claims 1 to 3, characterized in that: The step of inputting the irradiance data into the weather state prediction model on the previous day to obtain the weather state prediction result on the target forecast day includes: Based on the irradiance data, using the irradiance short-term prediction model, the irradiance prediction result of the target prediction day is obtained; Based on the irradiance prediction result, the weather state classification and recognition model is used to obtain the weather state prediction result of the target prediction day.

5. The method according to claim 4, characterized in that Before inputting the irradiance data into the weather state prediction model on the day before and obtaining the weather state prediction result on the target forecast day, the method further includes: Receive historical daily irradiance data; Based on the historical daily irradiance data, determining a corresponding daily weather state label; A weather state prediction model is trained based on the weather state label and the corresponding irradiance data.

6. A photovoltaic power generation short-term prediction device, characterized in that: The device comprises: A data acquisition module is used to obtain photovoltaic power data and irradiance data of the day before the target forecast day; A sequence acquisition module, used to input the photovoltaic power data into the first model and the second model respectively; obtain a first prediction sequence and a second prediction sequence of photovoltaic power; A weather prediction module, used to input the irradiance data into a weather state prediction model on the previous day to obtain a weather state prediction result on the target prediction day; The sequence fusion module is used to select a target fusion mode based on the weather state prediction result, fuse the first prediction sequence with the second prediction sequence, and obtain a short-term prediction result of photovoltaic power generation.

7. The device according to claim 6, characterized in that The device also includes: A model selection module, where the first model is a LSTM-RNN model; The second model includes a reference matrix constructed at a specific prediction time based on historical photovoltaic power data; the second model is used to predict the power generation on the target prediction day based on the data of the same date and adjacent dates in history; The reference value calculation module is also used to obtain a periodicity scale parameter and a similarity scale parameter; the periodicity scale parameter is used to indicate the number of years of historical data selected forward from the prediction time point; the similarity scale parameter is used to indicate the number of days of adjacent historical data selected from the prediction time point; Based on the periodicity scale parameter, the similarity scale parameter and the historical photovoltaic power data, a reference matrix is ​​constructed; The reference value is calculated using the reference matrix, and the reference value represents the photovoltaic power characteristics of the same date and adjacent dates in history.

8. The device according to claim 6 or 7, characterized in that The device also includes: An irradiance prediction module is used to obtain an irradiance prediction result for a target prediction day based on the irradiance data using an irradiance short-term prediction model; Based on the irradiance prediction result, using the weather state classification and recognition model, obtain the weather state prediction result of the target prediction day; Model training module, used to receive historical daily irradiance data; Based on the historical daily irradiance data, determining a corresponding daily weather state label; A weather state prediction model is trained based on the weather state label and the corresponding irradiance data.

9. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method according to any one of claims 1 to 5 by executing the computer instructions.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the method according to any one of claims 1 to 5.

Images